BEGIN:VCALENDAR
VERSION:2.0
X-WR-CALNAME:osskorea2026
X-WR-CALDESC:Event Calendar
METHOD:PUBLISH
CALSCALE:GREGORIAN
PRODID:-//Sched.com Open Source Summit Korea 2026//EN
X-WR-TIMEZONE:UTC
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260810T223000Z
DTEND:20260811T094500Z
SUMMARY:Registration + Badge Pick-up
DESCRIPTION:\n
CATEGORIES:REGISTRATION / BREAKS / SPECIAL EVENTS
LOCATION:Grand Ballroom Foyer\, Seoul\, South Korea
SEQUENCE:0
UID:bda0cedba5ffb5fa74ac04dd79680b83
URL:http://osskorea2026.sched.com/event/bda0cedba5ffb5fa74ac04dd79680b83
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260810T223000Z
DTEND:20260811T094500Z
SUMMARY:Sponsor Showcase
DESCRIPTION:\n
CATEGORIES:SPONSOR SHOWCASE
LOCATION:Sponsor Showcase - Grand Ballroom Foyer\, Seoul\, South Korea
SEQUENCE:0
UID:080b2807a5d3f3555c96c11de3816c84
URL:http://osskorea2026.sched.com/event/080b2807a5d3f3555c96c11de3816c84
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T000000Z
DTEND:20260811T002500Z
SUMMARY:Keynote: Welcome + Opening Remarks - Jim Zemlin\, CEO\, The Linux Foundation
DESCRIPTION:\n
CATEGORIES:KEYNOTE SESSIONS
LOCATION:Grand Ballroom 2-3\, Seoul\, South Korea
SEQUENCE:0
UID:5990783318c01281c9c2fe4ef52f1e25
URL:http://osskorea2026.sched.com/event/5990783318c01281c9c2fe4ef52f1e25
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T003000Z
DTEND:20260811T011500Z
SUMMARY:Keynote Sessions To Be Announced
DESCRIPTION:\n
CATEGORIES:KEYNOTE SESSIONS
LOCATION:Grand Ballroom 2-3\, Seoul\, South Korea
SEQUENCE:0
UID:83119cf6be8d60937aa552c7529b167b
URL:http://osskorea2026.sched.com/event/83119cf6be8d60937aa552c7529b167b
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T011500Z
DTEND:20260811T013000Z
SUMMARY:Keynote: Priya Nagpurkar\, Vice President\, Hybrid Cloud and AI Platforms\, IBM Research
DESCRIPTION:\n
CATEGORIES:KEYNOTE SESSIONS
LOCATION:Grand Ballroom 2-3\, Seoul\, South Korea
SEQUENCE:0
UID:a0854f70e0627a24d20082f9f5c9542b
URL:http://osskorea2026.sched.com/event/a0854f70e0627a24d20082f9f5c9542b
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T013000Z
DTEND:20260811T020000Z
SUMMARY:Morning Break
DESCRIPTION:\n
CATEGORIES:REGISTRATION / BREAKS / SPECIAL EVENTS
LOCATION:Grand Ballroom Foyer\, Seoul\, South Korea
SEQUENCE:0
UID:ce481f155898e87db1f642f93a0bf45c
URL:http://osskorea2026.sched.com/event/ce481f155898e87db1f642f93a0bf45c
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T020000Z
DTEND:20260811T023000Z
SUMMARY:Building Event-Driven WebAssembly on Kubernetes: Runtime\, Observability\, and Security - Brandon Kang\, Akamai Technologies & Nam Hai\, Hylatek JSC
DESCRIPTION:As cloud native systems evolve\, WebAssembly(WASM) is emerging as a runtime that complements and sometimes challenges container based approaches. With fast startup\, strong isolation\, and portability\, WASM enables efficient and secure event driven workloads. \n \n In this session\, we explore how to build and run the systems using WebAssembly on Kubernetes\, focusing on SpinKube\, an open source project that brings WASM workloads into Kubernetes. \n \n Through a live demonstration\, we will deploy and run WASM workloads using SpinKube\, highlighting real workflows. We will also compare cold start and execution performance\, showing millisecond startup and lower latency than traditional containers. Plus to demonstrate observability using eBPF based telemetry\, we enable visibility into runtime behavior without overhead. Attendees will learn how to trace execution\, monitor performance\, and troubleshoot distributed WASM systems. \n \n Finally\, we will talk how WebAssembly improves security and operational efficiency by reducing the attack surface and enabling consistent execution across environments\, providing a practical guide for production adoption.
CATEGORIES:CLOUD + ORCHESTRATION
LOCATION:Orchid 2\, Seoul\, South Korea
SEQUENCE:0
UID:2558b1dae1e099d9c8370d974106e86e
URL:http://osskorea2026.sched.com/event/2558b1dae1e099d9c8370d974106e86e
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T020000Z
DTEND:20260811T023000Z
SUMMARY:Upstream Kernel Hardening: Recent Progress and Challenges - Gustavo A. R. Silva\, Linux Kernel Self-Protection Project
DESCRIPTION:With the release of Linux 7.0\, Rust is no longer considered experimental. New and safer components are expected to be written in Rust in the near future. However\, the Linux kernel still contains more than 35 million lines of code written in C. This code will remain critical for years to come and\, for the benefit of everyone\, must continue to be maintained\, improved\, and hardened where possible. \n \n In the Linux Kernel Self-Protection Project\, we care deeply about this\, and we've been advancing kernel hardening upstream for several years. In this presentation\, we'll review recent hardening efforts and discuss the challenges we continue to face as we work toward our ultimate goal of eliminating classes of bugs and methods of exploitation in the upstream Linux kernel.
CATEGORIES:LINUX
LOCATION:Orchid 1\, Seoul\, South Korea
SEQUENCE:0
UID:273fb5c7d48d5ed617b45ffa32aeaf9f
URL:http://osskorea2026.sched.com/event/273fb5c7d48d5ed617b45ffa32aeaf9f
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T020000Z
DTEND:20260811T023000Z
SUMMARY:Squeezing Every Millisecond: A Practical Guide To Optimizing Time To First Token With OSS Muscle - Hrittik Roy\, Platform Advocate
DESCRIPTION:Large language models are getting faster GPUs every year\, yet users still notice the pause before the first word appears. That pause has a name: Time To First Token (TTFT). And in production LLM systems\, shaving even a few hundred milliseconds from it can dramatically change how responsive an application feels. \n \n This talk tells the story of where those milliseconds go. \n \n We will walk through the lifecycle of a request in modern LLM serving systems and explore the practical techniques engineers use to reduce TTFT in real deployments. Using examples from open source stacks like vLLM\, TensorRT-LLM\, and Hugging Face TGI\, we will examine four powerful optimization levers: KV cache strategies\, speculative decoding\, model quantization\, and batching policies. \n \n Instead of focusing only on theory\, the session highlights the tradeoffs practitioners face. When does speculative decoding actually help? When does batching hurt latency? When does quantization reduce memory pressure enough to speed up the first token? \n \n Attendees will leave with a practical playbook for diagnosing TTFT bottlenecks and choosing the right optimization strategy for their model\, infrastructure\, and workload.
CATEGORIES:OPEN AI + DATA
LOCATION:Grand Ballroom 2-3\, Seoul\, South Korea
SEQUENCE:0
UID:8dd38cf6ac8a56dce3c1598f06713e1a
URL:http://osskorea2026.sched.com/event/8dd38cf6ac8a56dce3c1598f06713e1a
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T020000Z
DTEND:20260811T023000Z
SUMMARY:SBOMs Aren't Enough. Secure Your Software Supply Chain End-To-End - Yongjae Chung\, New York University Secure Systems Lab & Justin Cappos\, New York University
DESCRIPTION:You probably heard that SBOMs are helpful\, but did you know that an SBOM only addresses a fraction of what can go wrong in your software supply chain? The SLSA (Supply Chain Levels for Software Artifacts) specification identifies 9 distinct threat areas\, spanning from source code\, all the way to package distribution. Most development teams address one or two of these and call it a day\, leaving gaps that real-world attacks like SolarWinds and Log4J have already exploited. We understand that it is difficult to cover all aspects when it comes to the software supply chain. \n \n How about we make this much easier? In this talk\, we will present an overview of the modern software supply chain threat model\, and show how you can provide integrity throughout the whole process of your software development life cycle. We will introduce an easy-to-setup\, end-to-end open source stack\, built from frameworks and tools within the CNCF/OpenSSF ecosystem.
CATEGORIES:PACKAGES + IMAGES + CONTAINERS
LOCATION:Chrysanthemum\, Seoul\, South Korea
SEQUENCE:0
UID:9c1cf6c56583e3475f610c3e861097ec
URL:http://osskorea2026.sched.com/event/9c1cf6c56583e3475f610c3e861097ec
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T024000Z
DTEND:20260811T031000Z
SUMMARY:Self-Healing Rollouts: Automating Production Fixes With Agentic AI - Kevin Dubois\, IBM & Carlos Sanchez\, Adobe
DESCRIPTION:Your software rollouts to production are probably always flawless\, right? For the rest of us\, once in a while we do run into issues when releasing code to production. Even with robust CI/CD\, production rollouts can hit unexpected snags. While in Kubernetes\, ArgoCD and Argo Rollouts excels at Progressive Delivery and automated rollbacks to mitigate deployment issues\, what if we could go a step further? \n \n This session explores how to elevate your release process by integrating Agentic AI and asynchronous coding agents\, with Argo Rollouts canary deployments. We'll demonstrate how an intelligent agent can automatically analyze a rollout failure\, pinpointing the root cause. Beyond diagnosis\, these agents can take proactive steps on your behalf\, suggesting and even implementing code fixes as new pull requests\, which can be redeployed automatically after PR review. This approach moves us closer to truly self-healing deployments. \n \n Join us to learn how to combine the power of cloud native projects like Kubernetes\, ArgoCD and Argo Rollouts with the autonomous capabilities of Agentic AI\, achieving a release experience that is not only seamless but also resilient.
CATEGORIES:CLOUD + ORCHESTRATION
LOCATION:Orchid 2\, Seoul\, South Korea
SEQUENCE:0
UID:bc40a44c43824e59ad49cfbaaac5cc04
URL:http://osskorea2026.sched.com/event/bc40a44c43824e59ad49cfbaaac5cc04
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T024000Z
DTEND:20260811T031000Z
SUMMARY:Getting Started With Kernel Programming: The Linux Kernel Mentorship Program - Jori Koolstra\, Independent
DESCRIPTION:The Linux Kernel is one of the most influential and widely-used open source software projects today. However\, getting started -- understanding the code and contributing to it -- can be quite daunting. Kernel code often carries history and conventions that are not always documented\, but are rather shared by the community as tribal knowledge. Moreover\, understanding operating-system-level source code comes with its own unique challenges: there is no libc\, you need to know about low-level primitives like memory barriers and various types of locking mechanisms\, and memory handling errors can bring the whole system down (no garbage collection either\, of course!). \n \n The Linux Kernel Mentorship Program (LKMP) addresses some of these difficulties and aims to mentor prospective contributors to overcome the initial hurdle -- because everything becomes easier after your first contribution. The program is led by experienced kernel maintainers with the assistance of regular contributors and is open to anyone. \n \n In this talk I will detail my experience as an LKMP mentee and explain how I have progressed since. I will also explain how you can get started with a small patch!
CATEGORIES:LINUX
LOCATION:Orchid 1\, Seoul\, South Korea
SEQUENCE:0
UID:67ec6f54d5e358011777e5d393b3780f
URL:http://osskorea2026.sched.com/event/67ec6f54d5e358011777e5d393b3780f
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T024000Z
DTEND:20260811T031000Z
SUMMARY:Stop Trusting a Black Box: The Economic Case for Open\, Sovereign AI - Vincent Caldeira\, Red Hat
DESCRIPTION:As AI matures from a novelty into a strategic asset\, 79% of organisations now prioritise sovereignty to mitigate vendor lock-in and secure critical data. Despite this\, the market remains paralysed by a paradox: while open models have achieved performance parity with proprietary systems at a fraction of the cost\, they remain massively under-utilised due to perceived friction and trust gaps. \n \n This session dissects this market inefficiency\, with insights from LF Research revealing how closed-source dominance is often driven by inertia rather than superior capability. We will explore how to operationalise true independence by rejecting "open-washing" in favour of rigorous frameworks that demand full model completeness: verifying everything from training data to weights. Join us to learn how to transition from renting black-box APIs to architecting a transparent\, reproducible\, and economically superior AI stack.
CATEGORIES:OPEN AI + DATA
LOCATION:Grand Ballroom 2-3\, Seoul\, South Korea
SEQUENCE:0
UID:16067da654fb22808f249e80b044e113
URL:http://osskorea2026.sched.com/event/16067da654fb22808f249e80b044e113
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T024000Z
DTEND:20260811T031000Z
SUMMARY:One Binary\, Every Package Manager: Shipping a Rust CLI To PyPI\, Npm\, Homebrew\, Winget\, and Beyond - Ajit Kumar\, Independent
DESCRIPTION:Most dev tools die in obscurity because installation friction kills adoption before the first command is run. This talk provides a battle-tested playbook for solving that problem using evnx—a Rust CLI for validating and secret-scanning `.env "files"—as a real-world case study. \n \n Launched in early 2026\, evnx achieved thousands of cross-ecosystem downloads within weeks by treating distribution as a first-class engineering concern. The session breaks down a complete distribution matrix: \n \n Registry Packaging: Using crates.io as a source of truth\, Maturin for native Python wheels\, and npm wrappers for platform-specific binaries. \n OS Package Managers: Automating Homebrew formulas via cargo-dist\, plus Scoop and Winget submissions. \n Developer Integration: GitHub Actions with SARIF output\, pre-commit hooks\, and lightweight Docker CI images. \n Supply-Chain Security: Leveraging PyPI Trusted Publishing (OIDC)\, provenance attestations\, and cosign for signed containers. \n \n Attendees will receive reusable GitHub Actions workflows and manifest templates from the public evnx repo to ship any Rust CLI across ecosystems without sacrificing security or maintainer sanity.
CATEGORIES:PACKAGES + IMAGES + CONTAINERS
LOCATION:Chrysanthemum\, Seoul\, South Korea
SEQUENCE:0
UID:1b94a72937f5b9ce664cbd106d4f2095
URL:http://osskorea2026.sched.com/event/1b94a72937f5b9ce664cbd106d4f2095
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T031000Z
DTEND:20260811T043500Z
SUMMARY:Lunch
DESCRIPTION:\n
CATEGORIES:REGISTRATION / BREAKS / SPECIAL EVENTS
LOCATION:Grand Ballroom Foyer\, Seoul\, South Korea
SEQUENCE:0
UID:3d547039abda157a677982fab46d4f7e
URL:http://osskorea2026.sched.com/event/3d547039abda157a677982fab46d4f7e
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T043500Z
DTEND:20260811T050500Z
SUMMARY:Beyond Round-Robin: GPU-Aware Load Balancing for LLM Inference in Kubernetes - Seokhwan Kong\, NETLOX
DESCRIPTION:Standard load balancers route LLM requests without awareness of KV-cache state or GPU queue depth — causing inflated Time-To-First-Token and wasted accelerator capacity.\n \n loxilb closes this gap with an eBPF-native AI gateway. The L4 layer uses XDP/TC and kernel sockmap for zero-copy forwarding. The L7 layer adds API-key validation\, token-quota enforcement\, and accelerator-aware routing via Consistent Hashing with Bounded Loads (CHWBL).\n \n This talk focuses on KV-exact routing for P/D disaggregated deployments. When requests share a prompt prefix\, prefill GPUs recompute identical KV tensors repeatedly — wasting up to 80% of GPU cycles. Tier 1.5 eliminates this: loxilb tokenizes the prompt in-process (HuggingFace Rust tokenizer via CGO)\, computes block hashes matching vLLM's internal format\, and routes to the exact GPU already holding those KV blocks — via a live inventory fed by vLLM's native ZMQ event stream.\n \n Unlike serving-layer schedulers (llm-d\, Dynamo)\, this runs in the eBPF data-plane hot path with no Kubernetes dependency — works on bare metal\, VMs\, and BlueField DPUs.\n \n We'll trace a live request through the P/D testbed and share lessons from building GPU-state-aware routing
CATEGORIES:CLOUD + ORCHESTRATION
LOCATION:Orchid 2\, Seoul\, South Korea
SEQUENCE:0
UID:f1c24589232e291093e556b633561189
URL:http://osskorea2026.sched.com/event/f1c24589232e291093e556b633561189
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T043500Z
DTEND:20260811T050500Z
SUMMARY:KNOD: In-Kernel Network Offload Device for GPU-Accelerated Packet Processing - Taehee Yoo\, Rebellions & Hoyeon Lee\, SUSE
DESCRIPTION:GPU compute resources remain inaccessible to the Linux kernel without userspace runtimes (ROCm\, CUDA)\, forcing network packet processing to stay CPU-bound regardless of available GPU capacity. \n We present KNOD (in-Kernel Network Offload Device)\, a Linux kernel framework that enables GPU-accelerated network packet processing entirely within the kernel. KNOD compiles and launches GPU kernels natively\, dispatching packet processing workloads directly to the GPU — no ROCm\, no userspace involvement\, using only publicly documented hardware interfaces. \n XDP offload is KNOD's first use case. Building on our LPC 2025 talk\, this talk covers: in-kernel GPU kernel compilation targeting AMD GFX9/Vega ISA with post-processing optimizations (latency: ~118µs → ~46µs)\; a lock-free GPU-side queue between NIC RX path and GPU compute\; and branch divergence mitigation to maximize SIMT utilization across heterogeneous packet flows. \n At equivalent throughput (~32 Mpps)\, KNOD reduces system-wide CPU utilization from 40% (CPU-based XDP) to 17-20%. We discuss lessons from MACsec/WireGuard offload attempts and the path toward KNOD as a general kernel framework for GPU-accelerated network functions.
CATEGORIES:LINUX
LOCATION:Orchid 1\, Seoul\, South Korea
SEQUENCE:0
UID:c16571dccded0017e0b7d3503886da33
URL:http://osskorea2026.sched.com/event/c16571dccded0017e0b7d3503886da33
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T043500Z
DTEND:20260811T050500Z
SUMMARY:Optimizing ML Inference Across Heterogeneous Accelerators - Sanjiban Sengupta\, CERN\, University of Manchester
DESCRIPTION:Machine learning is central to high-energy physics\,.These environments impose strict latency\, throughput\, and memory constraints\, requiring data processing at unprecedented rates. Addressing these demands requires efficient\, hardware-aware inference across heterogeneous architectures. \n \n The ML4EP team at CERN is developing an integrated ecosystem for this purpose. We present recent advances in efficient ML deployment. First\, aie4ml ports trained models to next-generation AMD FPGAs for ultra-low-latency inference. PQuantML complements this with hardware-aware pruning and quantization\, reducing model size and compute cost while preserving performance. \n \n For CPU and GPU inference\, SOFIE translates trained models into optimized C++ for heterogeneous systems. Using alpaka\, it enables backend-agnostic execution while minimizing data movement. It integrates with PQuantML to support quantized models and is deployable in both online systems and offline workflows.
CATEGORIES:OPEN AI + DATA
LOCATION:Grand Ballroom 2-3\, Seoul\, South Korea
SEQUENCE:0
UID:2776d941f45e9df86dcc77df1719f035
URL:http://osskorea2026.sched.com/event/2776d941f45e9df86dcc77df1719f035
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T043500Z
DTEND:20260811T050500Z
SUMMARY:Panel: Realizing Sovereign AI: Strategies for Korea’s Tech Sovereignty and AI Independence Via Open Source - Yongkook Kim\, IBM; Hong-Seok Kim\, Rebellions; Rosa (Hyun Kyong) Lee\, Korea Information Society Development Institute & Carlos Costa\, IBM Research
DESCRIPTION:When global AI development being centralized around proprietary "black-box" models\, the demand for Sovereign AI has become a national priority for many countries\, including South Korea. True sovereignty requires more than just local data or local LLMs\; it demands independence across the entire stack—from silicon up to the software services\, as well as AI model itself. This panel challenges the misconception that global technology leaders are incompatible with national goals\, demonstrating instead how open-source collaboration is the only viable path to technical and data independence as foundation for Sovereign AI.
CATEGORIES:OSS ENABLING & MANAGEMENT
LOCATION:Chrysanthemum\, Seoul\, South Korea
SEQUENCE:0
UID:202304193bb84170a2c12db0a0c812b4
URL:http://osskorea2026.sched.com/event/202304193bb84170a2c12db0a0c812b4
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T051500Z
DTEND:20260811T054500Z
SUMMARY:How I Tricked ArgoCD Into Sharding on a Single Cluster - Faeka Ansari\, Slice Financial Bank
DESCRIPTION:One controller at 846m CPU. The other at 6m. Both running. Both supposedly doing the same job. \n \n We had two ArgoCD application controllers. Scaling looked solved on paper. Except -- one was doing all the work and the other was just... sitting there. \n \n The thing nobody tells you when you first set up ArgoCD HA is that sharding works at the cluster level\, not the application level (sadly). So when you only have one cluster -- it doesn't matter how many controller replicas you spin up. The work doesn't split. It all piles onto one pod. \n \n We were running 300+ apps across 11 EKS clusters in a banking environment. Dropping availability was not an option. So we went digging. \n \n What we found was not in the official docs. It was a trick - something so counterintuitive that the first time I thought of it\, I laughed. Then I tried it. Then it worked. \n \n This talk is the story of that fix -- how we found it\, why it works\, and how we shipped it safely to production using Terraform without touching a single app config.
CATEGORIES:CLOUD + ORCHESTRATION
LOCATION:Orchid 2\, Seoul\, South Korea
SEQUENCE:0
UID:aae598acbd6407815fd2ed1a0cce49d0
URL:http://osskorea2026.sched.com/event/aae598acbd6407815fd2ed1a0cce49d0
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T051500Z
DTEND:20260811T054500Z
SUMMARY:Advanced Performance Profiling Using Perf Tools - Namhyung Kim\, Google
DESCRIPTION:Explore advanced Linux perf techniques for sophisticated performance analysis. This session delves into specialized areas beyond typical CPU profiling: \n \n * Data Type Profiling: Learn how to associate memory access hotspots with specific data structures and fields\, gaining insights into data layout efficiency. \n * Latency Profiling: Discover methods to identify and analyze serial execution segments within parallel programs that act as scalability bottlenecks. \n * Kernel Lock Contention profiling: Understand how to use perf to detect and analyze kernel-level lock contention issues. \n * And more.. \n \n This talk will cover the concepts and perf tool methodologies to diagnose these critical performance aspects in both user-space and kernel contexts\, empowering you to pinpoint elusive performance problems.
CATEGORIES:LINUX
LOCATION:Orchid 1\, Seoul\, South Korea
SEQUENCE:0
UID:1e4faac52d2759bf085001d8b7397b7b
URL:http://osskorea2026.sched.com/event/1e4faac52d2759bf085001d8b7397b7b
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T051500Z
DTEND:20260811T054500Z
SUMMARY:The Era of Vibe Coding: Why High-Skill Engineers Are More Critical Than Ever - Yongjin Lee\, Songnae High-school
DESCRIPTION:"Let’s be real: My AI has dementia." \n "Vibe-coding in real: Me and My AI is stupid" \n Welcome to the "Vibe Coding" era\, where anyone can prompt their way to complex code. But as a 17-year-old developer who "bullied" LLMs to build a custom Linux Based OS (MaruxOS) from scratch\, I’ve seen the ugly truth: AI is a brilliant assistant\, but a chronic liar with a 5-minute memory. \n In this session\, I expose the messy reality of AI-native development. I’ll share how I battled "Contextual Dementia"—where AI forgets my ARM64 architecture mid-patch—and survived "Mindless Yes-Clicking Syndrome\," where a single trusted prompt almost nuked my entire glibc. \n Key Takeaways: \n The Hallucination Hunter: Sniffing out AI’s "confident bullshit." \n The Dementia Doctor: Managing AI’s memory loss to maintain architectural integrity. \n The Final Decider: Moving beyond the "Yes-Clicking" trap to take true technical ownership. \n AI might be writing the code\, but only a real engineer can stop it from burning the house down.
CATEGORIES:OPEN AI + DATA
LOCATION:Grand Ballroom 2-3\, Seoul\, South Korea
SEQUENCE:0
UID:37158f4364a53890de454eed4a5a9f4a
URL:http://osskorea2026.sched.com/event/37158f4364a53890de454eed4a5a9f4a
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T051500Z
DTEND:20260811T054500Z
SUMMARY:From Contribution To Culture: 14 Years of Building an OSPO That Outgrew Itself - Darae Ahn\, Samsung Electronics
DESCRIPTION:Over the past decade\, many organizations have established OSPOs to manage open source usage and compliance. However\, building a sustainable open source culture requires more than policies and processes. \n \n This session shares a 14-year journey of an OSPO that evolved from a contribution-focused group into a broader organization encompassing usage\, compliance\, and internal enablement. \n \n It explores how open source practices were embedded into engineering culture through project incubation\, developer engagement\, and internal leadership programs. Over time\, these efforts led to a shift where open source activities became self-sustaining\, with teams proactively initiating projects and contributions. \n \n The session also reflects on an unexpected outcome: talent mobility. As internal open source leaders grew\, many moved on to new opportunities\, revealing both retention challenges and the broader impact of cultivating open talent. \n \n Key lessons include the balance between control and autonomy\, the role of leadership in cultural change\, and how open source can be viewed not only as a compliance requirement\, but as a long-term investment in culture and organizational brand.
CATEGORIES:OSS ENABLING & MANAGEMENT
LOCATION:Chrysanthemum\, Seoul\, South Korea
SEQUENCE:0
UID:96733595db85c49aebc9c51387e3fd11
URL:http://osskorea2026.sched.com/event/96733595db85c49aebc9c51387e3fd11
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T055500Z
DTEND:20260811T062500Z
SUMMARY:The Butterfly Effect of a Broken Disk: Top-Down Ceph Troubleshooting To Upstream Contribution - Sangyun Lee\, CJ Olivenetworks
DESCRIPTION:Tech blogs usually talk about huge Ceph clusters with thousands of disks. But in reality\, many of us run smaller on-prem setups. I will share my real experience of debugging a small Ceph cluster (10 nodes\, 10 NVMe\, 15 normal SSDs) and how tracking a slow app led me to write an upstream C++ patch. \n \n It started when our Valkey (Redis) pods suffered from severe write latency. We checked CephFS metadata and Istio network metrics\, but they were fine. So we dug into the storage layer using ceph osd perf. We saw huge latency on one specific node. Looking at the kernel logs (dmesg -k)\, we found a failing NVMe disk. I will explain the "Slow OSD" issue—how one broken disk can freeze a 3-replication cluster. \n \n During this outage\, reading ceph osd perf was very frustrating because the OSD IDs were completely unordered. Since it made debugging harder\, I decided to fix it. I looked into Ceph's C++ code\, changed the unordered hash map to a sorted vector (std::sort)\, and opened PR #67915 (https://github.com/ceph/ceph/pull/67915). I will share my experience discussing the fix with Ceph maintainers and why I believe engineers should fix the open-source tools they use.
CATEGORIES:CLOUD + ORCHESTRATION
LOCATION:Orchid 2\, Seoul\, South Korea
SEQUENCE:0
UID:15ee15e467ca4586a436844c2d1d329f
URL:http://osskorea2026.sched.com/event/15ee15e467ca4586a436844c2d1d329f
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T055500Z
DTEND:20260811T062500Z
SUMMARY:Bridge X86-64 Applications To ARM64 and RISC-V With Dynamic Binary Translation - Jim Huang & Chi-Kuan Chiu\, National Cheng Kung University
DESCRIPTION:Running unmodified x86-64 binaries on ARM64 and RISC-V Linux systems remains important\, but existing solutions often involve high overhead or depend on platform-specific hardware. Box64 addresses this gap with a wrapping-first architecture that prioritizes native execution by routing calls into host libraries through ABI translation\, while reserving JIT compilation for guest code that cannot be directly mapped. \n \n This session presents the core design of Box64 in three parts. First\, a native wrapping layer that spans hundreds of libraries\, handling entry-point interception and callback bridging. Second\, a multi-pass JIT compiler that applies optimizations such as flag liveness analysis and deferred flag computation. Third\, system-level integration techniques that enable practical deployment\, including container bypass\, mixed-bitness coordination with Wine\, and memory ordering support on weakly ordered architectures. \n \n We will also share performance results on ARM64 and RISC-V platforms\, showing how Box64 achieves near-native performance in favorable cases and significant speedups over traditional emulation approaches.
CATEGORIES:LINUX
LOCATION:Orchid 1\, Seoul\, South Korea
SEQUENCE:0
UID:f340c111e50b2dffc1d8cc1c30407960
URL:http://osskorea2026.sched.com/event/f340c111e50b2dffc1d8cc1c30407960
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T055500Z
DTEND:20260811T062500Z
SUMMARY:Open Source AI Agents on User-Owned Infra: K3s\, MCP\, and GPU Sharing in Practice - Miley Fu\, OlaresOS
DESCRIPTION:Most open-source AI demos stop at a chatbot\, but real users want agents that can work with private files\, call tools\, and run close to their important data without sending everything to the public cloud. \n \n This talk shows a practical reference architecture for private agent workflows on a single-node K3s-based personal cloud\, including local model serving\, private knowledge bases\, app sandboxing\, secure remote access\, and multiple GPU allocation modes for competing AI workloads. \n \n Using an open-source personal cloud OS as a case study\, I will share what works\, what breaks\, and which design choices matter when you try to make local AI usable by developers\, creators\, and small teams rather than just homelab experts.
CATEGORIES:OPEN AI + DATA
LOCATION:Grand Ballroom 2-3\, Seoul\, South Korea
SEQUENCE:0
UID:390900b91ebcd9bb7caf9905dda657ae
URL:http://osskorea2026.sched.com/event/390900b91ebcd9bb7caf9905dda657ae
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T055500Z
DTEND:20260811T062500Z
SUMMARY:How AI Is Changing Open Source Communities: Lessons From OpenEuler - Jianmin Wang\, openEuler Community
DESCRIPTION:Artificial Intelligence is reshaping how software is developed and maintained. From code generation to automated reviews\, AI tools are increasingly influencing how open source communities collaborate. This also introduces new challenges\, including how to handle AI-generated contributions\, maintain trust and code quality\, and define governance for AI-assisted workflows. \n \n In this session\, we share experiences from the openEuler community in integrating AI into development processes\, including AI-assisted code review\, package maintenance\, and community guidelines for AI usage\, as well as work on frameworks such as Intelligence BooM. \n \n We will discuss how these changes affect contributor workflows\, what challenges maintainers face in practice\, and what approaches have worked so far. The goal is to provide practical reference points for other open source communities exploring similar directions.
CATEGORIES:OSS ENABLING & MANAGEMENT
LOCATION:Chrysanthemum\, Seoul\, South Korea
SEQUENCE:0
UID:595aa617f0fb0656e8433babd05b6f67
URL:http://osskorea2026.sched.com/event/595aa617f0fb0656e8433babd05b6f67
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T062500Z
DTEND:20260811T065500Z
SUMMARY:Afternoon Break
DESCRIPTION:\n
CATEGORIES:REGISTRATION / BREAKS / SPECIAL EVENTS
LOCATION:Grand Ballroom Foyer\, Seoul\, South Korea
SEQUENCE:0
UID:6ac01daa7960db249f018e49f16eabf7
URL:http://osskorea2026.sched.com/event/6ac01daa7960db249f018e49f16eabf7
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T065500Z
DTEND:20260811T072500Z
SUMMARY:From Region To Multi-AZ: Building Resilient Cloud Infrastructure With OpenStack\, Kubernetes\, and OVN - 승진 한\, kt cloud
DESCRIPTION:As cloud service providers evolve their infrastructure\, Multi-AZ architecture becomes essential for service continuity\, failure isolation\, and operational resilience. However\, building a Multi-AZ cloud with open source technologies is not simply about spreading components across data centers. It requires design decisions across compute\, networking\, storage\, observability\, automation\, and failure validation. \n \n This session shares a CSP’s journey from a region-centric architecture toward a Multi-AZ cloud model using open source technologies such as OpenStack\, Kubernetes\, OVN\, Kube-OVN\, OpenStack-Helm\, Cluster API\, Ceph\, and cloud native observability tools. \n \n The talk will cover regional versus AZ-local service design\, OpenStack control plane deployment on Kubernetes\, zone-aware networking\, storage replication\, image availability\, observability\, and DR validation. It will also discuss trade-offs in traffic locality\, failover behavior\, data consistency\, and complexity. \n \n Rather than presenting a vendor-specific platform\, this session focuses on reusable architecture patterns and lessons for cloud operators who want to build resilient open cloud infrastructure with open source software.
CATEGORIES:CLOUD + ORCHESTRATION
LOCATION:Orchid 2\, Seoul\, South Korea
SEQUENCE:0
UID:1e2e7ca193b58b40d54a9cfa7b8de4ba
URL:http://osskorea2026.sched.com/event/1e2e7ca193b58b40d54a9cfa7b8de4ba
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T065500Z
DTEND:20260811T072500Z
SUMMARY:Maximizing Heterogeneous Memory Bandwidth in Multi-Socket Systems - Rakie Kim\, SK Hynix; Yunjeong Mun & Honggyu Kim\, SK hynix
DESCRIPTION:The advent of heterogeneous memory like CXL allows systems to secure additional bandwidth\, but effectively utilizing it remains challenging. While widely adopted memory tiering optimizes latency by migrating hot pages to fast memory\, bandwidth-intensive workloads require utilizing multiple memory tiers simultaneously. \n \n To address this\, Weighted Interleaving was introduced to distribute pages proportionally based on each node's bandwidth. However\, it has a critical limitation: it ignores the physical topology of multi-socket systems\, often leading to inefficient cross-socket memory accesses. \n \n To resolve this\, we introduce Socket-aware Weighted Interleave\, an advanced memory policy currently being upstreamed to the mainline Linux kernel. It recognizes multi-socket boundaries\, preventing performance degradation from unintended remote accesses and maximizing total throughput via localized allocation. \n \n This advanced policy is a newly integrated feature of HMSDK\, an open-source toolkit introduced last year. This session will focus primarily on this new capability\, exploring how HMSDK efficiently manages large-scale CXL and heterogeneous memory systems.
CATEGORIES:LINUX
LOCATION:Orchid 1\, Seoul\, South Korea
SEQUENCE:0
UID:76aaf38471b3a25a3d1fe0386658bf8c
URL:http://osskorea2026.sched.com/event/76aaf38471b3a25a3d1fe0386658bf8c
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T065500Z
DTEND:20260811T072500Z
SUMMARY:MCP Adoption and Why OSPO Skills Matter - Ana Jiménez Santamaría\, Linux Foundation (TODO Group and PyTorch Foundation)
DESCRIPTION:As organizations explore the Model Context Protocol (MCP)\, many of the early conversations naturally happen around AI tooling\, integrations\, and experimentation. At the same time\, MCP may also raise questions that are familiar to Open Source Program Offices (OSPOs) and related teams or professionals with open source management experience\, including governance\, contribution strategy\, standards engagement\, and cross-functional coordination \n \n This talk reflects on what organizations might gain by bringing those perspectives into MCP discussions earlier\, as well as into their broader Agentic AI strategy. Rather than treating MCP only as a technical AI topic\, the session will explore it as an area where open source process knowledge may also add value. Attendees will leave with ideas for how OSPOs can contribute to MCP adoption in practical\, collaborative\, and organization-aware ways
CATEGORIES:OPEN AI + DATA
LOCATION:Grand Ballroom 2-3\, Seoul\, South Korea
SEQUENCE:0
UID:9f7e7b6f8dd0b988cad36eff4acd2562
URL:http://osskorea2026.sched.com/event/9f7e7b6f8dd0b988cad36eff4acd2562
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T065500Z
DTEND:20260811T072500Z
SUMMARY:Skills-as-Packages: A Package Manager for AI Agent Skills - Brahada Srinivas\, Amazon
DESCRIPTION:AI agents like Claude Code\, Cursor\, and Codex learn libraries via SKILL.md files\, but these skills are currently unversioned\, ungoverned\, and unshared. We solved code dependency management with pip and npm — now it's time to solve it for AI knowledge. \n This talk presents an open-source\, package-manager-style system for agent skills. Skills are linked to their packages\, versioned with semver\, declared in skills.toml\, and locked via skills-lock.toml — just like regular dependencies. \n The CLI (skills add\, install\, lock\, publish) feels native to any developer using pip or uv. \n We'll cover: \n \n The SKILL.md open standard (YAML frontmatter + Markdown) — model-agnostic and runtime-agnostic \n Manifest format supporting version constraints\, inheritance\, and monorepo scoping \n Resolver that enforces constraint narrowing across org hierarchies \n Registry with publishing\, discovery\, approval workflows\, and security scanning \n Real cases where this prevented production incidents by keeping agents on correct\, up-to-date patterns \n \n Live demo: Add a skill\, resolve dependencies\, publish it\, and watch a new engineer's agent instantly get the right knowledge - no onboarding docs required.
CATEGORIES:OPEN AI + DATA
LOCATION:Chrysanthemum\, Seoul\, South Korea
SEQUENCE:0
UID:96a5e447d0ea2930736ecb10b5f7c25a
URL:http://osskorea2026.sched.com/event/96a5e447d0ea2930736ecb10b5f7c25a
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T073500Z
DTEND:20260811T080500Z
SUMMARY:What Happens When Your AI Agent Meets OPA - Jyoti Bisht\, Harness
DESCRIPTION:Tom and Jerry has run for over 80 years. Every episode follows the same plot: Tom builds an elaborate trap\, Jerry walks straight through it\, the house is destroyed\, and the owner blames Tom. Sound familiar? \n \n This talk is structured exactly like a Tom and Jerry cartoon except Tom is OPA and Jerry is your AI agent. Jerry is not malicious. He just wants the cheese. He will find every gap in every policy\, squeeze through every webhook\, and retreat to his mouse hole (MCP) the moment Tom gets close. The audience will root for Jerry. Jerry is still the problem. \n \n We walk through three episodes. Episode one: Jerry discovers he can call kubectl delete and Tom's first policy stops him — but not before he's renamed two deployments. Episode two: Jerry finds a namespace Tom forgot to cover and provisions a GPU node at $8/hour. Episode three: Jerry and Tom finally cooperate — the agent runs a legitimate right-sizing workflow\, OPA approves every step\, and the cluster is actually better for it. \n \n \n You'll leave with an OPA policy library for agentic tool-call governance\, Argo Workflows\, and a threat model built from real agent misbehaviour.
CATEGORIES:CLOUD + ORCHESTRATION
LOCATION:Orchid 2\, Seoul\, South Korea
SEQUENCE:0
UID:1b8e9894a60c655009521bfebaedd423
URL:http://osskorea2026.sched.com/event/1b8e9894a60c655009521bfebaedd423
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T073500Z
DTEND:20260811T080500Z
SUMMARY:Flexible Paging in Linux: Per-process Page Size - Dev Jain\, Arm Ltd
DESCRIPTION:Applications often run faster on a 64K page size kernel than on 4K. This is because larger pages reduce TLB pressure and pagetable walk overhead\, greatly improving memory access speed. But\, sysadmins are hesitant to use a 16K/64K kernel as most applications have been historically tuned for 4K pages\, and larger pages can increase memory waste due to internal fragmentation. This creates a tradeoff between performance and memory waste. \n \n We propose a design that changes the game by separating the page size of the user and the kernel. Instead of enforcing a single system-wide page size\, processes can operate under different page-size ABIs. A performance-critical app can use a 64K page-size ABI while still running on a 4K kernel. We achieve this by enabling Linux to provide memory in 64K chunks via 16 contiguous 4K pages\, and by translating operations on the 4K page table into a 64K “native” page table that serves as the actual hardware page table. \n \n At the same time\, lightweight or memory-sensitive applications can continue using 4K pages to minimize memory waste. All of this happens on the same Linux kernel\, without requiring separate builds of the kernel.
CATEGORIES:LINUX
LOCATION:Orchid 1\, Seoul\, South Korea
SEQUENCE:0
UID:36979b9b9792da161dd5d0b01985f226
URL:http://osskorea2026.sched.com/event/36979b9b9792da161dd5d0b01985f226
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T073500Z
DTEND:20260811T080500Z
SUMMARY:Benchmarking Beyond OpenSearch: Multi-Engine Vector Search Performance With OSB - Mike Oviedo\, AWS
DESCRIPTION:As vector search becomes foundational to AI workloads\, teams need reliable ways to evaluate engine performance before committing to one. We extended OpenSearch Benchmark (OSB) to support Milvus and Vespa alongside OpenSearch\, making it possible to compare throughput\, latency\, and recall across engines using the same datasets and query patterns. \n \n In this talk\, we'll walk through how OSB's new engine-as-module architecture lets you benchmark any search engine with a single CLI command. We'll show how customizable workload parameters control everything from HNSW graph construction to bulk ingestion strategies\, and demonstrate how to visualize comparative results in OpenSearch Dashboards\, turning raw benchmark data into actionable performance insights. \n \n Whether you're evaluating engines for a new project or optimizing an existing deployment\, you'll leave with practical knowledge of how to run your own benchmarks and how to extend OSB to support additional engines by implementing five functions.
CATEGORIES:OPEN AI + DATA
LOCATION:Grand Ballroom 2-3\, Seoul\, South Korea
SEQUENCE:0
UID:46612eaddbd936f7838f620ac60f0c5d
URL:http://osskorea2026.sched.com/event/46612eaddbd936f7838f620ac60f0c5d
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T073500Z
DTEND:20260811T080500Z
SUMMARY:Computer Programming Is Dead; Long Live AI-First Programming - Stephen Chin\, Neo4j & Cassandra Chin\, Independent
DESCRIPTION:Computer science graduates are facing an increasingly difficult job market. Recent data shows a sharp decline in employment outcomes for computer science majors\, highlighting the mismatch between what universities teach and what employers now demand. The traditional model of teaching syntax first and hoping students eventually build something useful is no longer working. In this keynote we argue that programming as we knew it is effectively dead. The future lies in AI-First programming\, built on the simple loop of try\, learn\, and grow. Learners try building code with AI assistance\, learn by unpacking the generated code and asking AI for detailed explanations\, and grow by testing and extending real applications. This loop not only builds confidence but also ensures we grow the generation of AI engineers that companies are desperate to hire.
CATEGORIES:OPEN AI + DATA
LOCATION:Chrysanthemum\, Seoul\, South Korea
SEQUENCE:0
UID:7e5fa4d1b75518af674c111b6a0d0d26
URL:http://osskorea2026.sched.com/event/7e5fa4d1b75518af674c111b6a0d0d26
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T081500Z
DTEND:20260811T084500Z
SUMMARY:From Frankenstein To Kamaji: Lessons in Building a Single CAPI Cluster Across Multiple Providers - Antonia von den Driesch\, Giant Swarm
DESCRIPTION:At Giant Swarm we use Cluster API to provision and bootstrap our k8s clusters. With this setup\, control plane (CP) and worker nodes must run on the same infrastructure which was never an issue so far... \n \n However\, in bare-metal environments\, using 128-core servers for CP nodes is luxury. It's far more efficient to host them as virtual machines on a hypervisor while keeping workers on physical hardware. But can we get around CAPI's limitations? \n \n We will walk through how we built Frankenstein's cluster by mixing vSphere for the CP and Proxmox for workers as a testing ground. While technically functional\, this required "hacky engineering". We will share the hurdles we hit and the operational risks of this hybrid cluster setup. \n \n Finally\, we will demonstrate how we solved this challenge with a cleaner\, upstream-friendly alternative. Kamaji lets us run the CP as pods in a management cluster. We achieved even better resource optimisation with full native community support and no custom hacks.
CATEGORIES:CLOUD + ORCHESTRATION
LOCATION:Orchid 2\, Seoul\, South Korea
SEQUENCE:0
UID:8367b881ed778ee43f826978c1ae5e15
URL:http://osskorea2026.sched.com/event/8367b881ed778ee43f826978c1ae5e15
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T081500Z
DTEND:20260811T084500Z
SUMMARY:Implementing a Stateless V4L2 Decoder Driver: Architecture and Lessons Learned - SungHo Lee\, Chips&Media
DESCRIPTION:Stateless video decoding has become the standard approach in V4L2 using the Request API\, but implementing a stateless decoder driver remains complex. It requires careful handling of codec parameters\, buffer management\, and coordination between userspace and the kernel. \n \n This session presents a practical overview of implementing a stateless V4L2 decoder driver based on real development experience. It covers how SPS/PPS and per-frame parameters are passed from userspace\, how reference frames are managed\, and how hardware constraints are handled within a generic V4L2 framework. \n \n We also discuss common pitfalls such as synchronization issues and buffer lifecycle management\, and highlight key differences from traditional stateful decoders. \n \n Attendees will gain a clear understanding of stateless decoder architecture and practical insights for developing robust V4L2 drivers.
CATEGORIES:LINUX
LOCATION:Orchid 1\, Seoul\, South Korea
SEQUENCE:0
UID:13881324a075f1666825cde7aee0c618
URL:http://osskorea2026.sched.com/event/13881324a075f1666825cde7aee0c618
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T081500Z
DTEND:20260811T084500Z
SUMMARY:GitAIOps: A 4-Layer Architecture for Predictable AI-Assisted Operations - Hoon Jo\, Megazone
DESCRIPTION:AI agents have no memory between sessions. Every conversation starts from zero. Git becomes the only persistent memory an AI agent can rely on. GitAIOps is the pattern built on this principle: Git is the memory\, and a 4-layer architecture defines what goes into that memory. \n \n I applied this to a production migration: 15 Helm releases\, Kafka ZooKeeper-to-KRaft\, Redis-to-Valkey\, full observability stack rebuild. The question: what does Git need to contain so any AI session picks up where the last one left off? \n \n The answer is a 4-layer Git structure\, each layer born from a production failure. \n Layer 1: Human plans in Git (36 files\, 23\,854 lines). Too verbose for AI. \n Layer 2: Distilled AI context in Git (6 files\, 1\,254 lines). 19:1 compression as a project state dashboard. \n Layer 3: Command Guardrails in Git (117 files). Enforced ordering\, no AI-generated commands. \n Layer 4: Locked values in Git (30 files). Zero interpretation\, reviewed like code. \n \n Every AI action reads from Git\, executes\, and commits back. The loop is closed. \n \n DEV: 2 weeks → 2 days. PROD: 1 week → 1 day. The session covers the architecture\, each layer's failure\, and real production artifacts.
CATEGORIES:OPEN AI + DATA
LOCATION:Chrysanthemum\, Seoul\, South Korea
SEQUENCE:0
UID:cd66b7f6d86397fdbb1297b9d6b9675f
URL:http://osskorea2026.sched.com/event/cd66b7f6d86397fdbb1297b9d6b9675f
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T081500Z
DTEND:20260811T084500Z
SUMMARY:Personalisation and Specialisation of Search With OpenSearch Agentic Search - Cedric Pelvet & Aswath Srinivasan\, AWS
DESCRIPTION:Agentic search lets you ask in Natural Language and have OpenSearch plan and execute retrieval. The Query Planner re-writes Natural Language to OpenSearch DSL using SOTA LLMs. This works amazingly well but not without important limitations: \n \n 1/ The biggest factor is added latency for remote inference. eCommerce Search demands sub-100ms response times. Even the best results when slow lead to search abandonment. \n 2/ We'll demo overcoming latency by hosting SLMs locally within OpenSearch nodes using vLLM/Ollama. This is faster & cheaper\, but SLMs suffer quality decline in query re-writes. We'll explore fine-tuning with domain-specific data—proving fine-tuned SLMs beat SOTA LLMs. \n 3/ Relevance remains generic even with Agentic Search. We'll show how to improve it using user and business contexts with hybrid search and reranking. \n \n This hands-on talk covers: \n 1/ Search Latency with Remote Inference \n 2/ Locally hosting SLMs using vLLM/Ollama \n 3/ Agentic search with SLM Local Inference \n 4/ Improving relevance with user and business contexts \n 5/ Fine-tuning SLMs for domain-specific query re-writing via Axolotl/LLaMA-Factory
CATEGORIES:OPEN AI + DATA
LOCATION:Grand Ballroom 2-3\, Seoul\, South Korea
SEQUENCE:0
UID:44a3983b5a6e6c01f54b0defaf9c9974
URL:http://osskorea2026.sched.com/event/44a3983b5a6e6c01f54b0defaf9c9974
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T084500Z
DTEND:20260811T094500Z
SUMMARY:Tux Trek
DESCRIPTION:\n
CATEGORIES:REGISTRATION / BREAKS / SPECIAL EVENTS
LOCATION:Sponsor Showcase - Grand Ballroom Foyer\, Seoul\, South Korea
SEQUENCE:0
UID:0be93238df9929c7cc89bb8f4d6ab765
URL:http://osskorea2026.sched.com/event/0be93238df9929c7cc89bb8f4d6ab765
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T230000Z
DTEND:20260812T084500Z
SUMMARY:Registration + Badge Pick-up
DESCRIPTION:\n
CATEGORIES:REGISTRATION / BREAKS / SPECIAL EVENTS
LOCATION:Grand Ballroom Foyer\, Seoul\, South Korea
SEQUENCE:0
UID:d7abd14095477aa8ace78b5a3ee37d92
URL:http://osskorea2026.sched.com/event/d7abd14095477aa8ace78b5a3ee37d92
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260811T230000Z
DTEND:20260812T065500Z
SUMMARY:Sponsor Showcase
DESCRIPTION:\n
CATEGORIES:SPONSOR SHOWCASE
LOCATION:Sponsor Showcase - Grand Ballroom Foyer\, Seoul\, South Korea
SEQUENCE:0
UID:ccd0d0de8502704bb1eac3c8d5bf9315
URL:http://osskorea2026.sched.com/event/ccd0d0de8502704bb1eac3c8d5bf9315
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T000000Z
DTEND:20260812T001000Z
SUMMARY:Keynote: Welcome Back - Jim Zemlin\, CEO\, The Linux Foundation
DESCRIPTION:\n
CATEGORIES:KEYNOTE SESSIONS
LOCATION:Grand Ballroom 2-3\, Seoul\, South Korea
SEQUENCE:0
UID:3b33818c1988b83d5c5b268a5a518352
URL:http://osskorea2026.sched.com/event/3b33818c1988b83d5c5b268a5a518352
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T001000Z
DTEND:20260812T010500Z
SUMMARY:Keynote Sessions To Be Announced
DESCRIPTION:\n
CATEGORIES:KEYNOTE SESSIONS
LOCATION:Grand Ballroom 2-3\, Seoul\, South Korea
SEQUENCE:0
UID:2e9ad9dd2e19eb60ddfab20d3fa341ed
URL:http://osskorea2026.sched.com/event/2e9ad9dd2e19eb60ddfab20d3fa341ed
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T011000Z
DTEND:20260812T012500Z
SUMMARY:Keynote: Building Trust in the Age of AI: From Data Openness to Responsible Innovation - Dr. Hongrak Lee\, President and Chief AI Officer\, LG AI Research
DESCRIPTION:\n\nThis presentation examines the evolving role of open data in the next phase of artificial intelligence development\, and the growing imperative to align innovation with trust. As AI systems become increasingly data-driven\, the quality\, origin\, and governance of data are emerging as foundational issues that will shape not only technological progress\, but also public confidence and global competitiveness.\n The talk situates these developments within a broader shift toward greater transparency and accountability in AI\, reflected in emerging policy directions such as the EU AI Act and Korea’s Korean AI Basic Act. Rather than viewing openness as a risk to be constrained\, it explores how responsible data practices and shared governance approaches can enable wider participation in AI development while maintaining integrity and trust.\n Ultimately\, the presentation argues that the future of AI will depend not only on model performance\, but on the strength of the underlying data ecosystem. Building sustainable trust around open data—through transparency\, collaboration\, and accountability—will be critical to unlocking the full potential of AI at scale.\n\n
CATEGORIES:KEYNOTE SESSIONS
LOCATION:Grand Ballroom 2-3\, Seoul\, South Korea
SEQUENCE:0
UID:733753d7f90747234987748ceddfd3f0
URL:http://osskorea2026.sched.com/event/733753d7f90747234987748ceddfd3f0
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T013000Z
DTEND:20260812T020000Z
SUMMARY:Morning Break
DESCRIPTION:\n
CATEGORIES:REGISTRATION / BREAKS / SPECIAL EVENTS
LOCATION:Grand Ballroom Foyer\, Seoul\, South Korea
SEQUENCE:0
UID:718757af1edaa3047e229cb03d6e7571
URL:http://osskorea2026.sched.com/event/718757af1edaa3047e229cb03d6e7571
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T020000Z
DTEND:20260812T023000Z
SUMMARY:Building and Orchestrating Production-ready Agentic AI Systems - Kevin Dubois\, IBM & Daniel Oh\, Red Hat
DESCRIPTION:Agentic AI is all the hype right now\, but how do you actually implement such a system for real enterprise\, cloud based use cases? \n \n The challenge for developers\, architects and platform engineers alike lies in custom building agents\, and even more so\, orchestrating these agents to collaborate effectively towards a common goal. Unfortunately though\, despite all the promises from vendors\, a "one-size-fits-all" or “off-the-shelf” approach just doesn't work due to the complex nature of software. In addition\, just like traditional apps\, these agentic systems will likely need to be deployed\, managed and observed in cloud environments. \n \n In this session we'll explore: \n * The spectrum of Agentic AI patterns \n * A real world-ish implementation of a highly performant - open source - agentic system (with Java!) \n * Deploying this agentic system to Kubernetes \n * Other considerations such as observability and fault tolerance to get it all running smoothly in production.
CATEGORIES:CLOUD + ORCHESTRATION
LOCATION:Orchid 2\, Seoul\, South Korea
SEQUENCE:0
UID:80f1cb5a3fe75dd22d089b12a846b7ca
URL:http://osskorea2026.sched.com/event/80f1cb5a3fe75dd22d089b12a846b7ca
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T020000Z
DTEND:20260812T023000Z
SUMMARY:Exploring Unikernel: An Empirical Comparison With Linux - Taekyung Kang & Kyungha Kim\, Boeing
DESCRIPTION:Unikernels are specialized operating systems designed for efficiency by including only the components needed by an application. By eliminating the traditional user–kernel separation and omitting general-purpose services such as background daemons and unused device drivers\, this design reduces system complexity and overhead while providing a fundamentally different execution model from conventional Linux systems. \n This study examines the structure of Unikraft-based unikernels from a library operating system perspective. It also presents an empirical comparison with Linux under controlled conditions. Both systems were deployed on the same Xen hypervisor and executed on a hardware platform\, running the same application with a common subset of POSIX APIs. Execution latency was measured\, and assembly-level analysis was performed to investigate potential reasons for the observed differences. \n The results are presented as an example of combining performance measurement and low-level inspection when analyzing specialized operating systems. Similar approaches may be applicable to other specialized OS or unikernel contexts\, depending on the application and execution environment.
CATEGORIES:EMBEDDED
LOCATION:Chrysanthemum\, Seoul\, South Korea
SEQUENCE:0
UID:5d257f5286a84c68c94889b7bf1fdbf9
URL:http://osskorea2026.sched.com/event/5d257f5286a84c68c94889b7bf1fdbf9
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T020000Z
DTEND:20260812T023000Z
SUMMARY:Fusing AOSP and Linux! To Open a New Chapter for Linux Desktops - Yong Gong\, Phytium: OpenFDE open-source community
DESCRIPTION:The Linux desktop system boasts strong customization and freedom\, but its application ecosystem has long been a weakness. Recently\, the Android system has millions of applications\, covering all aspects of work and life. If we could integrate these and create a brand-new desktop experience\, what kind of impact would it have? \n \n This presentation will introduce our open-source project - OpenFDE (Open Fusion Desktop Environment). This is an innovative desktop environment that integrates the Android open-source project (AOSP) deeply with Linux applications\, enabling users to run various Android applications seamlessly on their Linux desktops. \n \n We will share: \n 1、Project Vision: Why did we choose AOSP and Linux integration? What problems are we aiming to solve? \n 2、Core Architecture: How do we achieve the integration of these two major systems? In-depth analysis of the key technologies of OpenFDE in terms of graphics\, window management\, and application lifecycle. \n 3、Live Demonstration: Witness firsthand the smooth coexistence of Linux and the latest Android applications (including large-scale games) in the OpenFDE environment.
CATEGORIES:LINUX
LOCATION:Orchid 1\, Seoul\, South Korea
SEQUENCE:0
UID:20c92b6775c33d9867b77148c14dd5c5
URL:http://osskorea2026.sched.com/event/20c92b6775c33d9867b77148c14dd5c5
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T020000Z
DTEND:20260812T023000Z
SUMMARY:LLMs Change — Where Should Knowledge Live? (Lessons From SBOM) - Koji Annoura\, Annoura Office
DESCRIPTION:Large Language Models (LLMs) are widely used to search\, summarize\, and generate answers. They are powerful\, but results are not always stable and can be difficult to verify. \n \n Many systems use Retrieval-Augmented Generation (RAG) to connect LLMs with external data. This helps\, but does not fully solve the problem. The same question can produce different answers depending on the model or context\, making knowledge hard to manage. \n \n This raises a simple question: where should knowledge be managed? \n \n A similar issue exists in software supply chains. SBOM (Software Bill of Materials) makes components and relationships visible\, but is often treated as static data. \n \n Based on hands-on experience designing graph-based knowledge systems\, this session introduces a design approach: managing knowledge outside LLMs as structured\, persistent data. \n \n We discuss lessons from SBOM and how graph-based approaches—using technologies such as SQL/PGQ or GQL—can help understand relationships and trace changes. \n \n The focus is a practical way of thinking for more open and sustainable knowledge practices.
CATEGORIES:OPEN AI + DATA
LOCATION:Grand Ballroom 2-3\, Seoul\, South Korea
SEQUENCE:0
UID:80b08ebccfd1f9f7e2078e5f96b3e0b0
URL:http://osskorea2026.sched.com/event/80b08ebccfd1f9f7e2078e5f96b3e0b0
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T020000Z
DTEND:20260812T023000Z
SUMMARY:When Semantic Search Breaks: RAM Walls\, Silent Failures\, and the Architecture Decisions That Actual - Jeevan D C\, Entain
DESCRIPTION:The team shipped semantic search. It worked. Then came 100M vectors\, a RAM bill that tripled overnight\, filtered queries silently returning zero results\, and a proposal to "just add Pinecone alongside Postgres." \n \n This talk is for engineers who've already built the thing and now need to scale it without burning the budget or the team. We'll do the RAM math that should happen before any architecture decision\, walk through quantization strategies that deliver 32x compression with 95%+ recall\, show why filtered search is the #1 silent production failure\, and lay out when DiskANN\, hybrid BM25+vector search\, or a specialized vector DB actually makes sense — and when it doesn't. \n \n Live demos included. Decision matrices to take back to the team on Monday. No hype\, just trade-offs in plain English.
CATEGORIES:OPEN AI + DATA
LOCATION:Rose\, Seoul\, South Korea
SEQUENCE:0
UID:8e3e3763a864d204d781965b4e3b7d55
URL:http://osskorea2026.sched.com/event/8e3e3763a864d204d781965b4e3b7d55
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T024000Z
DTEND:20260812T031000Z
SUMMARY:From Static Rules To Reasoning Platforms: Scaling Intelligent Canary Delivery in 2026 - Daniel Oh\, Red Hat
DESCRIPTION:As organizations scale their Kubernetes footprint\, the "Day 2" reality of GitOps becomes clear: static thresholds are brittle. Standard Canary rollouts rely on fixed Prometheus queries (e.g.\, Error Rate &lt\; 1%)\, but these rules lack the context to distinguish between a minor transient blip and a systemic failure. For Platform Engineers\, this results in "Alert Fatigue" and manual "promotion" gates that slow down the delivery pipeline. \n In 2026\, we are moving from Static Automation to Reasoning Platforms. \n This session explores how to evolve your delivery infrastructure into an intelligent system that doesn't just follow rules\, but reasons through data. We will demonstrate how to wrap ArgoCD Rollouts with an Agentic Reasoning Layer capable of cross-referencing metrics\, logs\, and distributed traces to make autonomous "Go/No-Go" decisions. \n \n We will trigger a Canary deployment that passes basic health checks but introduces a "silent failure" (e.g.\, a cache hit-rate drop causing downstream latency). You will see the Reasoning Platform detect the anomaly\, pause the rollout\, "investigate" the root cause\, and present a natural-language justification for the automated rollback.
CATEGORIES:CLOUD + ORCHESTRATION
LOCATION:Orchid 2\, Seoul\, South Korea
SEQUENCE:0
UID:ad8749dcd7099c3a43ce19ac7aef1327
URL:http://osskorea2026.sched.com/event/ad8749dcd7099c3a43ce19ac7aef1327
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T024000Z
DTEND:20260812T031000Z
SUMMARY:From CVEs To Compliance: Automating Embedded Linux Kernel Security - Kyungsik Lee\, LG Electronics
DESCRIPTION:Global security regulations such as the EU Cyber Resilience Act (CRA) have raised security requirements for embedded products. Open source components\, especially the Linux kernel\, must now support systematic vulnerability management\, fast security patching\, and long-term maintenance\, making kernel security a key challenge. \n \n This session discusses practical solutions for managing Linux kernel vulnerabilities in embedded products. It begins with an overview of recent kernel CVE trends and their impact on long-lived and customized kernels. The session then introduces a CI-based vulnerability response pipeline designed to minimize the time from CVE disclosure to patch deployment. \n \n A key challenge is backporting security fixes to older or vendor-modified kernels\, where patches often do not apply cleanly. To address this\, the session presents an AI agent–based approach that assists developers by analyzing CVE data\, upstream patches\, and kernel context to suggest candidate backports. \n \n By adopting an AI-assisted vulnerability response workflow\, teams can reduce response time and prepare for compliance with evolving global security regulations.
CATEGORIES:EMBEDDED
LOCATION:Chrysanthemum\, Seoul\, South Korea
SEQUENCE:0
UID:02ab65c2112f081cd6fc3c4763c69da9
URL:http://osskorea2026.sched.com/event/02ab65c2112f081cd6fc3c4763c69da9
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T024000Z
DTEND:20260812T031000Z
SUMMARY:Run Queue Secrets: How the Linux Scheduler Shapes Your Application Performance - Anjali Jain\, Amazon Web Services & Abhineet Saxena\, Atlassian
DESCRIPTION:In modern cloud environments\, CPU usage often looks healthy while applications still experience latency and unpredictable slowdowns. In one Kubernetes based system running latency sensitive workloads\, services degraded despite CPU staying below 50%. Traditional monitoring showed no clear issue\, leading to delayed debugging and reactive fixes. \n \n This case study highlights how the real bottleneck was hidden in the Linux scheduler specifically run queue contention and scheduling delays. By examining run queue depth and task scheduling behavior\, the team uncovered how fairness-driven scheduling (CFS) impacted performance under load. The session also touches on how newer approaches like EEVDF aim to improve scheduling decisions. \n \n The result was faster root cause identification\, better workload tuning\, and improved application responsiveness without scaling resources. \n \n Key Takeaways: \n \n - Why CPU utilization can mislead performance analysis \n - How run queue depth and scheduling latency impact applications \n - The gap between fair scheduling and real-world performance \n - What’s changing in modern schedulers (CFS &gt\; EEVDF) \n - Practical ways to reason about scheduler related bottlenecks
CATEGORIES:LINUX
LOCATION:Orchid 1\, Seoul\, South Korea
SEQUENCE:0
UID:2831622f22ec9b1505f64851ee2b0ce8
URL:http://osskorea2026.sched.com/event/2831622f22ec9b1505f64851ee2b0ce8
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T024000Z
DTEND:20260812T031000Z
SUMMARY:From Black Box To Insight: Observability for AI Agents in Production - Mostafa Radwan\, Datadog & Brandon Kang\, Akamai Technologies
DESCRIPTION:AI agents are quickly moving from prototypes to production systems that manage APIs\, tools\, and complex reasoning tasks. \n \n Once these systems are deployed\, they often act like black boxes\, hiding failures\, slowdowns\, and unexpected behavior. \n \n This session will show you how to add observability to agentic systems with open-source tools and cloud-native tech. \n \n ​Brandon and Mostafa will break down the lifecycle of an AI agent\, covering prompt execution\, tool invocation\, memory access\, and multi-agent coordination\, and demonstrate how to make each stage observable. \n \n Using a practical\, architecture-focused approach\, they will show how to: \n \n - Trace agent workflows across distributed systems \n - Monitor latency and token usage \n - Detect anomalies such as hallucinations\, tool misuse\, and runaway behavior. \n - Connect LLM behavior with infrastructure metrics like GPU\, container\, and network data. \n \n They will also present a complete reference architecture that uses open-source projects including OpenTelemetry (OTel)\, Prometheus\, Grafana\, and some new tools for LLM observability. \n \n Attendees will learn practical ways to build transparent and reliable production AI agent systems.
CATEGORIES:OPEN AI + DATA
LOCATION:Grand Ballroom 2-3\, Seoul\, South Korea
SEQUENCE:0
UID:6efd109fce48f9fe7562f12317c5ab06
URL:http://osskorea2026.sched.com/event/6efd109fce48f9fe7562f12317c5ab06
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T024000Z
DTEND:20260812T031000Z
SUMMARY:Simplifying Edge Compute for Open-Source AI - Reza Jelveh\, Dynamia
DESCRIPTION:The open-source community is rapidly producing powerful\, autonomous AI agents capable of complex reasoning and tool use\, such as Hermes and OpenClaw. However\, the barrier to entry for deploying these agents remains high\, often requiring complex infrastructure setups and significant computational resources. To truly democratize access to agentic AI\, we must simplify the deployment and management of edge compute. \n This technical deep dive breaks down the complexities of orchestrating edge resources for autonomous agents. We will focus on practical strategies for deploying hybrid Kubernetes clusters designed specifically for local AI workloads. The session will highlight the critical role of GPU management\, demonstrating how to optimize hardware utilization through memory slicing and time-sharing with open source projects. By providing a blueprint for efficient edge orchestration\, this talk aims to empower developers and builders to deploy sophisticated open-source agents without the need for massive cloud budgets.
CATEGORIES:OPEN AI + DATA
LOCATION:Rose\, Seoul\, South Korea
SEQUENCE:0
UID:a02eae80832dcd98e03dd4ab3726b4ff
URL:http://osskorea2026.sched.com/event/a02eae80832dcd98e03dd4ab3726b4ff
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T031000Z
DTEND:20260812T043500Z
SUMMARY:Lunch
DESCRIPTION:\n
CATEGORIES:REGISTRATION / BREAKS / SPECIAL EVENTS
LOCATION:Grand Ballroom Foyer\, Seoul\, South Korea
SEQUENCE:0
UID:c1acaa0f86cc62ba2130747e58d18c55
URL:http://osskorea2026.sched.com/event/c1acaa0f86cc62ba2130747e58d18c55
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T043500Z
DTEND:20260812T050500Z
SUMMARY:EZIO: Predictable\, Fast\, Scalable BitTorrent-Based Bare Metal Provisioning - Date (Yu-Chiang) Huang\, DozenCloud
DESCRIPTION:Deploying OS images to bare metal clusters is painful. Unicast scales linearly with node count. Multicast stalls if one node is slow. Past BitTorrent approaches either transfer entire raw partitions (wasting bandwidth) or require RAM buffering for image conversion (size limited). \n \n EZIO's provisioning time depends on image size and bandwidth\, not node count. It transfers only used filesystem blocks and writes directly to raw disk by calculating offsets on the fly. No RAM buffering\, no image conversion\, no size limit. Each node works independently. Broken nodes can rejoin after recovery. This enables deploying large HPC environments with pre-installed software and data. Clonezilla has integrated EZIO for production use. \n \n Benchmarks: On HDD (50GB\, 32 nodes)\, 11x faster than unicast\, 50% faster than multicast. In the cluster with NVMe SSD and 10G network at Taiwan's National Center for High-performance Computing (NCHC)\, 500 MB/s across 32 nodes. Lab tests reach 700 MB/s. \n \n This talk covers EZIO's architecture\, real-world benchmarks\, and integration approach.
CATEGORIES:CLOUD + ORCHESTRATION
LOCATION:Orchid 2\, Seoul\, South Korea
SEQUENCE:0
UID:f8d2fd390fa99ac72027b271b27cdd53
URL:http://osskorea2026.sched.com/event/f8d2fd390fa99ac72027b271b27cdd53
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T043500Z
DTEND:20260812T050500Z
SUMMARY:From Closed To Collaborative: Perspectives and Lessons From Qualcomm’s Open Development Experience - Craig Northway\, Qualcomm Technologies Inc
DESCRIPTION:For more than 15 years\, Qualcomm’s been actively involved in a range of Open Source ecosystems. Until recently\, some parts of our development were handled behind closed doors\, with contributions coming a bit later and enablement being somewhat limited. We tried various projects and partnerships to push things upstream sooner\, but it wasn’t until lately that we truly made a complete shift. \n \n Over the past 18 months\, we’ve totally revisited our approach—moving an entire Linux product development ecosystem\, with hundreds of contributors\, from a private downstream setup to a full-blown Open Development model. This wasn’t just a surface change: it meant overhauling how our engineers work\, syncing up our internal systems with open practices\, and fundamentally changing the way our developers connect and collaborate. \n \n In this session\, we’ll share what made this transition work for us—including how we managed to weave our internal systems into Open Source workflows\, encouraged developers to embrace new ways of thinking\, and built scalable processes that can handle all sorts of Linux ecosystems and distributions.
CATEGORIES:EMBEDDED
LOCATION:Chrysanthemum\, Seoul\, South Korea
SEQUENCE:0
UID:3b5783e46fe82ac627b0ea2e2f47cd63
URL:http://osskorea2026.sched.com/event/3b5783e46fe82ac627b0ea2e2f47cd63
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T043500Z
DTEND:20260812T050500Z
SUMMARY:Kernel Live Patching: Mitigating CVEs With Zero Downtime - Shung-Hsi Yu\, SUSE
DESCRIPTION:Service operators often face a constant friction: maintaining high Service Level Indicators (SLIs) versus addressing immediate security vulnerabilities. While userspace updates cause minimal disruption\, mitigating Linux kernel vulnerabilities traditionally mandates a full system reboot\, forcing administrators into the dilemma of choosing between proactive security practices and continuous uptime. \n \n This session explores the workings of Linux kernel live patching\, details how a livepatch kernel module is built\, examines the internal mechanisms that power the technology (e.g.\, ftrace)\, and provides a practical overview of the current ecosystem so administrators can start using live patches immediately.
CATEGORIES:LINUX
LOCATION:Orchid 1\, Seoul\, South Korea
SEQUENCE:0
UID:712b211eb901515a4f7e8898331a209f
URL:http://osskorea2026.sched.com/event/712b211eb901515a4f7e8898331a209f
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T043500Z
DTEND:20260812T050500Z
SUMMARY:Prompt Injection Is the New SQL Injection: Securing Tool-Using Agents - Jigyasa Grover\, Uber & Rishabh Misra\, Atlassian
DESCRIPTION:As LLM-based agents gain access to tools - APIs\, databases\, file systems\, and internal services\, the security model changes. The model is no longer only generating text\; it is selecting actions and invoking capabilities across systems. \n \n Prompt injection attacks exploit this boundary. between model reasoning & external execution. \n \n This talk examines how tool-enabled agents built on open LLM frameworks expand the attack surface and why traditional input validation approaches are insufficient. \n \n We will analyze concrete failure modes such as: \n - Prompt injection vs classical injection: control of model reasoning rather than query structure \n - Tool outputs as secondary injection vector in multi-step workflows \n - Why system prompts & guardrails are not reliable isolation boundaries \n - Capability scoping & least-privilege design for tool access \n - Isolation patterns for tool execution (sandboxing\, mediated execution layers) \n - Structured tool interfaces vs free-form prompting \n - Observability patterns for tracing agent decisions and tool calls \n - Adversarial testing of agent pipelines \n \n Examples draw from patterns emerging in open-source LLM and agent ecosystems.
CATEGORIES:OPEN AI + DATA
LOCATION:Grand Ballroom 2-3\, Seoul\, South Korea
SEQUENCE:0
UID:c769483994c71379543e4612f7f8f129
URL:http://osskorea2026.sched.com/event/c769483994c71379543e4612f7f8f129
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T043500Z
DTEND:20260812T050500Z
SUMMARY:AI-Powered Open Source Risk Management: ISO Self-Certification Kit and 5-Level AI Coding Governance - Haksung Jang\, SK Telecom
DESCRIPTION:Trusted OSS is an open-source self-certification kit that guides any organization from zero to ISO/IEC 5230 (license compliance) and ISO/IEC 18974 (security assurance) conformance using AI agents. \n No prior expertise required. \n \n Built by the OpenChain Korea Work Group and released under CC BY 4.0\, the kit features: \n \n • AI agents (Claude Code) that auto-generate company-specific compliance artifacts: OSS policy\, SBOM\, vulnerability response procedures\, training curriculum\, and conformance declaration \n • DevSecOps pipelines (SAST\, SCA\, secret detection\, IaC) ready to drop into any CI/CD environment \n • A 5-level AI Coding Governance Maturity Model — from ad-hoc prompting (Level 1) to AI-augmented defense (findings-driven review\, AI fuzzing) and continuous auto-remediation (Level 5). \n \n In this session\, I'll walk through how any team can go from no compliance process to a fully documented\, self-certifiable program in hours\, not months. I'll also share how we're using AI to close the compliance skills gap across Korean enterprises and SMEs — making OpenChain certification accessible to all. \n \n Attendees leave with a working toolkit they can clone and run today.
CATEGORIES:OSS ENABLING & MANAGEMENT
LOCATION:Rose\, Seoul\, South Korea
SEQUENCE:0
UID:d379fddbd3a8831365e7a000c74f5df4
URL:http://osskorea2026.sched.com/event/d379fddbd3a8831365e7a000c74f5df4
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T051500Z
DTEND:20260812T054500Z
SUMMARY:Clouds on Clouds: OpenStack and Kubernetes With Cloud-Barista - Seokho Son\, ETRI
DESCRIPTION:Can you build OpenStack and Kubernetes clusters anywhere\, across multiple clouds\, and still understand how network connectivity works? \n \n This session explores that question through a scenario using Cloud-Barista\, an open-source multi-cloud orchestrator. \n \n Instead of separate project overviews\, this talk connects Cloud-Barista\, OpenStack\, and Kubernetes as infrastructure layers. Cloud-Barista provisions VMs on public clouds\, then OpenStack is deployed on them. The OpenStack-based cloud is registered back into Cloud-Barista to create VMs and host a web service. \n \n Will that service be reachable from Internet? If not\, why? What makes it complex? These questions guide our explanation of network paths and reachability. We will apply the same lens to Kubernetes on multi-cloud VMs. \n \n The focus is not just automation\, but how connectivity works: public/private IPs\, bastion access\, cluster nodes\, and how users reach workloads. \n \n Beginners curious about these topics will gain practical insight into open-source infrastructure stacks. \n \n This is not another cluster deployment talk. It is an experimental journey across open-source cloud layers\, from multi-cloud IaaS to OpenStack and Kubernetes.
CATEGORIES:CLOUD + ORCHESTRATION
LOCATION:Orchid 2\, Seoul\, South Korea
SEQUENCE:0
UID:2741c52442417f62467ad838e7108f34
URL:http://osskorea2026.sched.com/event/2741c52442417f62467ad838e7108f34
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T051500Z
DTEND:20260812T054500Z
SUMMARY:Finding Vulnerabilities in IoT Embedded Devices Using Linux OS and Open Source Tools - Dr. Nkuba Kayembe Carlos\, Korea University
DESCRIPTION:Smart home ecosystems are increasingly powered by embedded Linux platforms\, yet the security of their underlying firmware\, memory management\, and wireless communication stacks remains dangerously underexamined. This talk presents a systematic approach to vulnerability discovery in IoT embedded Z-Wave smart home devices using freely available Linux OS tools and developed open source frameworks — bridging the gap between theoretical security research and hands-on embedded testing. \n \n Drawing directly from original research that resulted in 18 CVEs assigned by U.S. CERT and U.S. MITRE\, and from a live-demonstration talk presented at TyphoonCon 2025 in Seoul\, the speaker will walk attendees through a structured open source testing methodology: \n \n • Fuzzing embedded protocol stacks to uncover memory-corruption vulnerabilities \n • Live exploitation: manipulating controller internal memory to delete or modify secured slave device properties \n • Triggering Denial-of-Service (DoS) conditions that disable an entire smart home network \n • Coordinated disclosure and remediation work with SiLabs and the Z-Wave Alliance \n
CATEGORIES:EMBEDDED
LOCATION:Chrysanthemum\, Seoul\, South Korea
SEQUENCE:0
UID:85be1c57ce7cdf55f6e898b6e6c7f422
URL:http://osskorea2026.sched.com/event/85be1c57ce7cdf55f6e898b6e6c7f422
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T051500Z
DTEND:20260812T054500Z
SUMMARY:Beyond Main(): Orchestrating Early Boot With Linker Scripts and ELF Sections - Antra Purohit & Hemant Bharadwaj\, Microsoft
DESCRIPTION:Modern Linux init systems don't just magically start executing services. Long before PID 1 reaches its main loop\, a complex\, carefully orchestrated dance of initialization routines takes place. This presentation explores the crucial\, often-overlooked machinery bridging the compiler toolchain and system orchestration: the linker. \n \n Often treated as a black box\, the linker (ld) is an immensely powerful tool for low-level systems engineers. This deep dive pulls back the curtain on how modern init architectures leverage custom linker scripts\, specifically defined ELF sections\, and compiler directives (like __attribute__((constructor))) to precisely control execution order and optimize memory layouts before the system fully comes alive.
CATEGORIES:LINUX
LOCATION:Orchid 1\, Seoul\, South Korea
SEQUENCE:0
UID:d1a50612c4ce8bb172b6811018a22546
URL:http://osskorea2026.sched.com/event/d1a50612c4ce8bb172b6811018a22546
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T051500Z
DTEND:20260812T054500Z
SUMMARY:How x402 Brings Open Source Governance to Payments - Junhyeok Yoo\, Four Pillars
DESCRIPTION:In April 2026\, x402 joined the Linux Foundation with over 20 founding members. It is the first payment protocol under open source governance at the Linux Foundation. \n \n x402 activates HTTP 402 "Payment Required\," reserved in 1997 but never used. A server responds with 402 and payment terms\, the client settles on-chain\, the resource is delivered. Apache 2.0\, zero protocol fees\, chain-agnostic\, no single-organization dependency. \n \n Part 1: Why now. Every internet layer runs on open protocols except payments. That breaks when AI agents must pay at machine speed. We map the agentic payment stack and ask: is it open\, who governs it? \n \n Part 2: How x402 works. HTTP 402 challenge-response\, stateless architecture\, one middleware line to gate any API\, production deployments today. \n \n Part 3: Bottom-up to foundation. x402 was open-sourced\, developers adopted it\, major infrastructure providers shipped native support\, competing protocols chose to integrate x402 rather than build a rival. Then it moved to the Linux Foundation. \n \n Part 4: What comes next. The protocol is early. Foundation governance changes the signal. We examine what sustained adoption requires.
CATEGORIES:OPEN AI + DATA
LOCATION:Grand Ballroom 2-3\, Seoul\, South Korea
SEQUENCE:0
UID:1362c4e57d6c347e86944e9bf114da0e
URL:http://osskorea2026.sched.com/event/1362c4e57d6c347e86944e9bf114da0e
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T051500Z
DTEND:20260812T054500Z
SUMMARY:How To Make AI-Assisted OSS Contributions Review-Ready - Jaewoo Choi\, Hyundai Autoever
DESCRIPTION:AI coding tools can generate patches quickly\, but maintainers still cannot merge a PR on generated code alone. They need evidence\, what was reproduced\, what changed\, how it was validated\, and whether the contributor actually reduced review work instead of shifting it downstream. \n \n In this session\, I draw on recurring patterns from reviewing and maintaining Argo CD contributions to show why many AI-assisted PRs stall and what makes others move quickly. I present a simple four-stage framework for review-ready OSS contributions\, Set up the environment\, validate the problem before changing code\, harden the fix with tests and checks\, and submit a PR with enough context and proof for efficient review. Attendees will leave with practical guidance they can use immediately to improve contribution quality and reduce maintainer burden.
CATEGORIES:OSS ENABLING & MANAGEMENT
LOCATION:Rose\, Seoul\, South Korea
SEQUENCE:0
UID:cfda5bb48285674a74e237e118fd3c03
URL:http://osskorea2026.sched.com/event/cfda5bb48285674a74e237e118fd3c03
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T055500Z
DTEND:20260812T062500Z
SUMMARY:Supply Chain Security in Air-Gapped Kubernetes: SBOM\, Provenance\, and What Breaks - Michel Schildmeijer\, SSC-ICT
DESCRIPTION:Running Kubernetes in air-gapped environments changes how the software supply chain behaves. Image distribution\, signature verification\, and dependency updates cannot rely on upstream access and need to be handled explicitly. \n This talk examines what breaks when enforcing SBOM and image provenance in restricted networks. It covers artifact promotion across trust boundaries\, signature verification without external services\, base image drift\, and coordinating updates across disconnected environments. \n The focus is on concrete failure patterns and trade-offs: broken trust chains\, stale dependencies\, inconsistent SBOM data\, and operational overhead introduced by manual controls. Several common supply chain practices do not translate directly to air-gapped setups. \n \n The session shows which parts of the supply chain need to be redesigned to keep provenance and integrity intact without relying on continuous connectivity to upstream ecosystems.
CATEGORIES:CLOUD + ORCHESTRATION
LOCATION:Orchid 2\, Seoul\, South Korea
SEQUENCE:0
UID:d726979c6ced724960a26ad73cabcc5e
URL:http://osskorea2026.sched.com/event/d726979c6ced724960a26ad73cabcc5e
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T055500Z
DTEND:20260812T062500Z
SUMMARY:From Continuous Tracing To Automated Insights: AI-Driven Performance Analysis With Guider - Peace Lee & Jaeguk Lee\, Hyundai Motor Company
DESCRIPTION:Catching transient performance drops is notoriously hard\; by the time you attach a profiler\, the bottleneck is often gone. Continuous tracing is the solution\, but manually analyzing the resulting massive data is a huge hurdle. \n \n This session introduces a practical methodology for always-on performance monitoring and automated analysis using Guider. First\, we demonstrate how Guider efficiently profiles system performance in the background with minimal overhead. Upon detecting an anomaly\, it automatically generates a retroactive report using trace data captured just before the event. \n \n We will then show how to feed these structured reports into an AI pipeline. Instead of manually deciphering complex kernel metrics\, you will see how AI instantly pinpoints the root cause of performance regressions. \n \n Attendees will learn to build a continuous monitoring pipeline that automatically turns raw trace data into real-time\, actionable insights.
CATEGORIES:LINUX
LOCATION:Orchid 1\, Seoul\, South Korea
SEQUENCE:0
UID:776afb3d06f305499bed7e491329d470
URL:http://osskorea2026.sched.com/event/776afb3d06f305499bed7e491329d470
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T055500Z
DTEND:20260812T062500Z
SUMMARY:Yukti: A Unified Inference Interface for Low-Latency Machine Learning in High-Energy Physics - Sanjiban Sengupta\, CERN\, University of Manchester
DESCRIPTION:Machine learning is increasingly used in high-energy physics\, particularly in trigger systems that process data at rates of 100 kHz while making real-time event selection decisions. The latency and reliability requirements demand highly optimized inference pipelines. While portable solutions such as ONNX Runtime simplify deployment\, many applications rely on hardware-specific libraries like NVIDIA TensorRT and MIGraphX (via ROCm) for optimal performance. Code-generation approaches such as SOFIE offer additional efficiency but introduce integration complexity. \n \n We present a unified inference interface that abstracts backend-specific details while preserving performance. It enables execution across multiple inference libraries without data copies or user-side configuration changes. An offline processor converts trained models into backend-optimized plans\, and a lightweight runtime loads and executes them through a common API with direct data access for heterogeneous environments.
CATEGORIES:OPEN AI + DATA
LOCATION:Grand Ballroom 2-3\, Seoul\, South Korea
SEQUENCE:0
UID:d75d6e7302a66bc1dd1369ae76e72943
URL:http://osskorea2026.sched.com/event/d75d6e7302a66bc1dd1369ae76e72943
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T055500Z
DTEND:20260812T062500Z
SUMMARY:Scaling Open Source Compliance With Argus - Luyen Vu\, SAP
DESCRIPTION:Managing open source compliance at scale is no longer a spreadsheet problem — it’s an infrastructure and intelligence challenge. This session introduces Argus\, an always-on compliance and code intelligence platform built for OSPOs and security teams managing hundreds of repositories. \n \n Argus unifies license scanning\, cryptographic analysis\, AI/LLM dependency detection\, infrastructure checks\, and coding agent artifact discovery into a single pipeline. Its hybrid architecture combines Python orchestration\, a high-throughput Rust engine for large repositories\, and an LLM-powered analysis layer that turns raw findings into actionable insights. \n \n We’ll demonstrate how techniques like blobless git cloning\, bounded concurrency\, and per-scanner task tracking enable scanning repositories with 100K+ files in minutes. Attendees will learn practical patterns for automated license resolution\, large-scale crypto detection\, and generating compliance reports that teams actually act on. \n \n Finally\, we’ll share lessons from integrating open source and commercial tools in a production OSPO — and how an agentic approach helps small teams scale compliance without increasing headcount.
CATEGORIES:OSS ENABLING & MANAGEMENT
LOCATION:Rose\, Seoul\, South Korea
SEQUENCE:0
UID:1c9615bdc0f67feed14c9749b8b73c55
URL:http://osskorea2026.sched.com/event/1c9615bdc0f67feed14c9749b8b73c55
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T055500Z
DTEND:20260812T062500Z
SUMMARY:Zephyr RTOS: 10 Years After Applying OSS Best Practices - Kate Stewart\, The Linux Foundation
DESCRIPTION:Zephyr initially set out to solve a problem that many embedded teams quietly struggled with: how to build dependable real-time systems without being locked into a single vendor\, toolchain\, or proprietary stack. The project introduced a new model built around portability\, adoption of open source and security best practices\, modern tooling\, and a shared ecosystem of drivers and middleware. \n \n From the start\, there was the commitment from the start apply known best practices to its development. While Zephyr is a different code base\, a lot of the lessons learned from developing the Linux Kernel were applied. The project has also focused on incorporating security best practices from the start which now enables it to make compliance easier for manufacturers looking to conform to the emerging Cybersecurity Resilence Act (CRA). \n \n Best practices have also enabled the project to work towards achieving formal safety certification for 61508 and 26262. The project has achieved 61508 concept approval at this point\, as is working towards formal certification\, using a combination of traditional V-Model analysis\, and innovative techniques to keep up with the speed of open source development.
CATEGORIES:ZEPHYR
LOCATION:Chrysanthemum\, Seoul\, South Korea
SEQUENCE:0
UID:df94eb0cab490b768c3739c5f214024f
URL:http://osskorea2026.sched.com/event/df94eb0cab490b768c3739c5f214024f
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T062500Z
DTEND:20260812T065500Z
SUMMARY:Afternoon Break
DESCRIPTION:\n
CATEGORIES:REGISTRATION / BREAKS / SPECIAL EVENTS
LOCATION:Grand Ballroom Foyer\, Seoul\, South Korea
SEQUENCE:0
UID:c74a8a42fae138fb7e2c21dee53b27f9
URL:http://osskorea2026.sched.com/event/c74a8a42fae138fb7e2c21dee53b27f9
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T065500Z
DTEND:20260812T072500Z
SUMMARY:Who Watches the Watchers? Building Observability for the Platform Itself Across Multi-Cluster EKS - Faeka Ansari\, Slice Financial Bank
DESCRIPTION:There is this moment every platform team hits where an alert fires at 1am\, everyone stares at it\, and nobody is quite sure what it means or whose job it is to fix it. \n \n That was us. 11 EKS clusters. 6 AWS accounts. Alerts routing to a channel. No runbooks. No context. Just noise. \n \n Here is what made it worse --- we were the team responsible for the observability stack itself. VictoriaMetrics\, vmagent\, vmselect\, Grafana\, CloudWatch --- we ran all of it. And most of it was set up just well enough to fire alerts\, but not well enough to actually help anyone during an incident. \n \n Most observability talks are about how to instrument your applications. This one is about what happens when the platform itself becomes the thing you need to observe! and you are the one responsible for both the problem and the solution. \n \n We will talk about what we got wrong first\, what a P1 at 1am actually teaches you about your own stack\, and what we built to make sure the next time something breaks\, we know exactly where to look within the first five minutes.
CATEGORIES:CLOUD + ORCHESTRATION
LOCATION:Orchid 2\, Seoul\, South Korea
SEQUENCE:0
UID:7ad1f7e1c4cf1edac336113b7e444270
URL:http://osskorea2026.sched.com/event/7ad1f7e1c4cf1edac336113b7e444270
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T065500Z
DTEND:20260812T072500Z
SUMMARY:Connecting the Dots With Context Graphs - Stephen Chin\, Neo4j
DESCRIPTION:AI systems need more than intelligence\; they need context that persists. Without it\, even strong models can misinterpret information\, lose decision rationale\, or repeat the same mistakes. Context Graphs have emerged as a practical pattern for agentic AI: a living graph that captures not only what was retrieved or known\, but how context led to actions through tool calls\, constraints\, policies\, and outcomes\, stitched across entities and time so precedent becomes searchable. \n \n This talk explores context engineering as the discipline of designing that context layer\, and shows how context graphs complement retrieval by enabling multi-hop\, structured context assembly (building on GraphRAG-style hierarchical summaries) while improving explainability and evaluation. Attendees will leave with a practical understanding of how to build context pipelines that combine contextual retrieval with persistent memory and provenance\, and why context graphs are becoming central to trustworthy\, enterprise-ready AI systems.
CATEGORIES:OPEN AI + DATA
LOCATION:Grand Ballroom 2-3\, Seoul\, South Korea
SEQUENCE:0
UID:17208c5a854add5b7718cb3f73935f8a
URL:http://osskorea2026.sched.com/event/17208c5a854add5b7718cb3f73935f8a
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T065500Z
DTEND:20260812T072500Z
SUMMARY:OpenChain’s Strategic Direction for 2027 in Global Practices - Meixia Wang\, The Linux Foundation
DESCRIPTION:In this session\, as the newly appointed Executive Director of OpenChain\, I will outline the strategic direction for OpenChain in 2027 and beyond. Building on the solid foundation of our existing ISO standards\, including ISO 5230:2020 and ISO 18974\, I will share the next steps OpenChain will take to expand its global reach. We will focus on fostering international collaboration\, adapting to emerging regulatory landscapes\, and driving innovation in open-source compliance practices worldwide.
CATEGORIES:OSS ENABLING & MANAGEMENT
LOCATION:Orchid 1\, Seoul\, South Korea
SEQUENCE:0
UID:a4b67558636f7bcacbe6c6418aabd95d
URL:http://osskorea2026.sched.com/event/a4b67558636f7bcacbe6c6418aabd95d
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T065500Z
DTEND:20260812T072500Z
SUMMARY:Case Studies of Existing Use of Linux in Safety-critical Domains - Nikita Verma\, Individual & Harshita Varma\, Independent
DESCRIPTION:The automotive transition to Software-Defined Vehicles (SDVs) relies on mixed-criticality architectures\, consolidating open-source infotainment (Automotive Grade Linux) alongside safety-critical Real-Time Operating Systems (RTOS). This virtualization boundary—often KVM/Xen—is assumed to be a secure airgap. However\, guest-to-host communication requires hardware abstraction\, primarily via the VirtIO standard. \n \n This 40-minute session conducts a hardcore technical teardown of the virtqueue shared-memory mechanism\, exposing how legacy C-based VirtIO backends (vhost-net) introduce critical vulnerabilities into the automotive supply chain. \n \n We will dissect a hypervisor escape utilizing custom fuzzing. By crafting malformed descriptor chains to bypass frontend validation\, a compromised guest can force the host's backend into out-of-bounds memory corruption\, effectively bridging the airgap into the control plane. \n \n Finally\, we will architect the open-source defense: migrating to memory-safe rust-vmm virtualization components to mathematically eliminate buffer overflows\, and deploying zero-overhead eBPF probes for kernel-level I/O anomaly detection.
CATEGORIES:SAFETY CRITICAL SOFTWARE
LOCATION:Chrysanthemum\, Seoul\, South Korea
SEQUENCE:0
UID:cf819d67cdedba4ba24f649e30cfcc37
URL:http://osskorea2026.sched.com/event/cf819d67cdedba4ba24f649e30cfcc37
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T073500Z
DTEND:20260812T080500Z
SUMMARY:Accelerating Open-Source 5G UPF on Kubernetes With eBPF (Live Demo) - Khushi Chhillar\, NgKore
DESCRIPTION:Running the 5G User Plane Function (UPF) as a Kubernetes pod using open-source stacks like free5GC often hits a wall: kernel networking overhead destroys packet throughput. eBPF can bypass this bottleneck\, but the concept remains intimidating to network engineers who don’t do kernel programming. \n \n This lightning talk provides a zero-to-understanding educational journey. The speaker will first illustrate\, with simple diagrams\, the kernel’s slow path versus the eBPF XDP/AF_XDP fast path for GTP-U packets. Then\, using a live (or pre-recorded) demo on a Minikube cluster\, they will show an open-source UPF accelerated by a small eBPF program—demonstrating how GTP encapsulation/decapsulation is handled in the driver\, with line-rate forwarding. The entire code\, including a ready-to-run Docker image and Helm chart\, will be shared on GitHub. Attendees will leave with a mental model of exactly where eBPF sits\, which hooks to use\, and how to evaluate eBPF acceleration for their own 5G cloud-native network functions. No prior eBPF or 5G core knowledge is required \,only curiosity about high-performance networking.
CATEGORIES:CLOUD + ORCHESTRATION
LOCATION:Orchid 2\, Seoul\, South Korea
SEQUENCE:0
UID:3cdbf3d7164c53038c05946df8f28e4c
URL:http://osskorea2026.sched.com/event/3cdbf3d7164c53038c05946df8f28e4c
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T073500Z
DTEND:20260812T080500Z
SUMMARY:Reboot Without Rebooting: Updating Linux Userspace at the Speed of a Service Restart - Nandakumar Raghavan & Prasanna Kumar T S M\, Microsoft
DESCRIPTION:What if you could replace your entire OS userspace — binaries\, configs\, services\, even init itself — without ever stopping the kernel? No firmware POST\, no hardware re-init\, no multi-minute downtime. Just a clean userspace restart in seconds. \n \n That's what soft-reboot (systemctl soft-reboot)\, implemented in systemd\, delivers: a primitive that tears down all of userspace\, pivots into a fresh root filesystem\, and re-launches PID 1 — while the running kernel stays untouched. \n \n This talk unpacks how soft-reboot works under the hood\, why it matters for modern OS update workflows\, and how it compares to a traditional reboot and kexec. We'll walk through real production use cases: zero-downtime OS updates\, staged fleet rollouts\, and rapid recovery from a wedged userspace — without paying the cost of a full boot cycle. \n \n Attendees will leave with a clear mental model of soft-reboot internals\, practical guidance on when to choose it over a full reboot or kexec\, and concrete patterns for building soft-reboot-aware update pipelines. \n \n If you've ever wished you could ship an OS update as fast as restarting a service\, this talk is for you.
CATEGORIES:LINUX
LOCATION:Orchid 1\, Seoul\, South Korea
SEQUENCE:0
UID:e56af631cc5e4525b53ac205e39c2eb1
URL:http://osskorea2026.sched.com/event/e56af631cc5e4525b53ac205e39c2eb1
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T073500Z
DTEND:20260812T080500Z
SUMMARY:Building Reliable AI Agents: An Open Source Approach To Evaluation and Observability - Sho Tanaka\, Snowflake
DESCRIPTION:Building AI agents is easier than ever with open source tools\, but ensuring their reliability in production remains a major challenge. Unlike traditional software\, AI agents are non-deterministic\, making simple pass/fail testing insufficient. \n \n This talk introduces a practical approach to evaluation and observability for AI agents\, combining open source tools such as TruLens with agent architectures inspired by AgentGPT. \n \n We will demonstrate how to instrument agent workflows\, capture execution traces\, and implement evaluation metrics such as faithfulness\, tool selection accuracy\, and answer relevancy. Attendees will also see how to visualize agent behavior and identify failure points across retrieval\, reasoning\, and generation layers using a lightweight dashboard. \n \n Finally\, we show how to build a feedback loop to iteratively improve agent performance\, and share a reference implementation (GitHub) that can be reused with different agent frameworks.
CATEGORIES:OPEN AI + DATA
LOCATION:Rose\, Seoul\, South Korea
SEQUENCE:0
UID:8ecf6a1728e63e8812a96b5ed607996d
URL:http://osskorea2026.sched.com/event/8ecf6a1728e63e8812a96b5ed607996d
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T073500Z
DTEND:20260812T080500Z
SUMMARY:Science in the Agentic Era: Structured Experimentation With Ado - Alessandro Pomponio & Michael Johnston\, IBM
DESCRIPTION:As AI agents become increasingly capable of generating code and executing complex workflows\, their use in research is still limited by concerns about rigour and reproducibility. This session introduces ado\, an open-source framework that brings structure to agent-driven scientific experimentation. \n \n ado defines schemas for the core elements of a discovery process: the problem space and how to explore it. Agents iteratively propose and refine experimental campaigns as validated configurations based on these schemas\, while ado handles execution. This separation of research intent from execution constrains agents to focus on the research task\, reducing hallucinations and the need to write boilerplate code. Combined with a set of agent skills for formulating problems\, creating and running experiments\, and analysing their results\, ado provides a framework for end-to-end agent-driven discovery workflows. \n \n Whether you are an experienced researcher or new to computational experimentation\, this talk presents a practical model for integrating AI agents into research workflows while keeping experimentation structured\, transparent\, and reproducible.
CATEGORIES:OPEN AI + DATA
LOCATION:Grand Ballroom 2-3\, Seoul\, South Korea
SEQUENCE:0
UID:10f26df60b0065c6b8ed41a59d54fc1b
URL:http://osskorea2026.sched.com/event/10f26df60b0065c6b8ed41a59d54fc1b
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T073500Z
DTEND:20260812T080500Z
SUMMARY:Using AI To Bridge the Gap Between Safety Standards and Open Source Development - Kate Stewart\, The Linux Foundation
DESCRIPTION:Popular open source operating systems like the Linux Kernel and Zephyr RTOS accept up to 9 commits per hour. Safety standards\, like 61508\, 26262\, and others were developed without this rate of change in mind. Safety standards also expect the requirements to be explicit\, which is not part of OS development processes. By using AI tools\, we're able to accelerate the analysis of OS code to derive the requirements and traceability to tests. By storing this info in tools that can import and export System Package Data eXchange (SPDX) 3.0+\, we're able to capture the requirements in a way that can be leveraged for wider system analysis necessary for safety. Associating integrity methods with the requirements and code snippets\, also enables monitoring. Combining requirements traceability with precise build SBOM metadata\, gives us a framework to keep a component compliant to a safety profile after a security fix. \n \n This talk will provide a view on the latest experiments occurring with the Linux Kernel in the ELISA project\, as well as in the Zephyr Safety Working group\, and SPDX Functional Safety working group to extend SPDX to meet the needs of establishing these frameworks.
CATEGORIES:SAFETY CRITICAL SOFTWARE
LOCATION:Chrysanthemum\, Seoul\, South Korea
SEQUENCE:0
UID:6f7602acfefc175d8aba592245b6d4ab
URL:http://osskorea2026.sched.com/event/6f7602acfefc175d8aba592245b6d4ab
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T081500Z
DTEND:20260812T084500Z
SUMMARY:Live Demo: Event-Driven Terraform Drift Detection With Falco - Keita Higaki\, Sysdig\,Inc
DESCRIPTION:Terraform is widely used to manage infrastructure as code\, but traditional drift detection relies on periodic scans or manual checks such as terraform plan. This approach often fails to detect real-time changes\, manual modifications\, or unauthorized actions. \n \n In this technical feature demonstration\, we present an open source approach to drift detection using an event-driven model powered by Falco. \n \n We will demonstrate how: \n \n infrastructure is provisioned using Terraform \n manual or out-of-band changes introduce drift \n Falco detects these changes in real time via event streams \n an open source tool analyzes and surfaces these events as actionable drift signals \n \n Unlike traditional drift detection tools\, this approach enables near real-time detection\, user attribution\, and continuous visibility into infrastructure changes. \n \n This session introduces the concept of “event-driven runtime drift” and shows how it complements Terraform-based workflows using open source technologies. \n \n The demo is based on a publicly available open source project\, allowing attendees to reproduce the setup and apply it to their own environments.
CATEGORIES:CLOUD + ORCHESTRATION
LOCATION:Orchid 2\, Seoul\, South Korea
SEQUENCE:0
UID:d197829bbadf8b824659e7923e4894fb
URL:http://osskorea2026.sched.com/event/d197829bbadf8b824659e7923e4894fb
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T081500Z
DTEND:20260812T084500Z
SUMMARY:Eval-Driven Development: Mastering Agentic Tracing and Expert-Aligned Judges With MLflow - Vincent Caldeira & Sharon Dashet\, Red Hat
DESCRIPTION:Building AI agents is easy\, but ensuring their reliability is hard. Because agents operate in open-ended\, non-deterministic environments\, traditional Test-Driven Development (TDD) falls short. The solution is Eval-Driven Development (EDD)\, a paradigm that embeds continuous evaluation throughout the agent lifecycle. \n \n This talk explores how to operationalize EDD. First\, we examine the critical role of observability. Evaluating just a final output is insufficient for multi-step workflows. We show how OpenTelemetry (OTel) compatible tracing exposes internal reasoning and tool usage for granular debugging. \n \n Next\, we tackle scaling evaluation via LLM-as-a-Judge. Since uncalibrated judges often miss domain-specific nuances\, we demonstrate aligning custom judges with human experts. Using open-source MLflow\, attendees will learn to capture traces\, collect expert feedback\, and use alignment optimizers to create expert-aligned evaluators.
CATEGORIES:OPEN AI + DATA
LOCATION:Rose\, Seoul\, South Korea
SEQUENCE:0
UID:703f744dfdbde0c84856328fe793b693
URL:http://osskorea2026.sched.com/event/703f744dfdbde0c84856328fe793b693
END:VEVENT
BEGIN:VEVENT
DTSTAMP:20260614T085700Z
DTSTART:20260812T081500Z
DTEND:20260812T084500Z
SUMMARY:Exploring HiFloat8: A Tapered Format Complementing the FP8 Ecosystem for Robust Model Training - Speakers To Be Announced
DESCRIPTION:Standard FP8 formats suffer from frequent gradient overflows and heavy reliance on complex Delayed Scaling\, which often lead to training instability or suboptimal convergence in large models. This session introduces HiFloat8 (HiF8) — a tapered precision format that offers an alternative approach to managing dynamic range. This "natural" alignment with neural network weight/gradient distributions allows HiF8 to capture high-magnitude outliers without the aggressive scaling required by standard FP8.We explore how HiF8 can works in the training and inference procedure.We will demonstrate the implementation of HiF8 within the ecosystem. They allow developers to evaluate performance of Hif8 on GPUs. Also\, we will give an analysis of training stability and final loss parity where HiF8 provides relatively the same accuracy and 1.5-1.7 times GEMM performance than FP16. Finally\, we will share insights from our ongoing collaboration on dedicated hardware support for HiF8.
CATEGORIES:OPEN AI + DATA
LOCATION:Grand Ballroom 2-3\, Seoul\, South Korea
SEQUENCE:0
UID:5c68885bf9c97b7e32e50e5dfe61e165
URL:http://osskorea2026.sched.com/event/5c68885bf9c97b7e32e50e5dfe61e165
END:VEVENT
END:VCALENDAR