Open Source Summit Korea 2025
2025-11-04..2025-11-05
Seoul, South Korea
[Open Source Summit Korea 2025, Open Source Summit Korea 2026]
who it's for
linux kernel maintainers, ospo managers, ai engineers, security engineers, cloud architects, embedded systems engineers, legal and compliance managers, platform engineers
topics
Wflex-array-member-not-at-end; RISC-V silicon and tooling; Rust in the Linux kernel; SBOM and VEX workflows; vector search with OpenSearch; LLM inference on edge with NNTrainer; orchestrating AI agents with Rust; supply chain security for AI; temporal partitioning in avionics; BPF-based double free detection; verified Debian packaging at scale
| title | author | length |
|---|---|---|
| Tutorial: Building Custom MCP Servers With FastMCP and Integrating AI Agents Using... | M. Emel | 50m |
Builds a working pipeline: a FastMCP server running on localhost:801 hosting tool endpoints, a Google ADK-based country agent with four tools (get country… | ||
| Keynote: Open Source × AI: Shaping the Future of kt cloud | Jaesuk Ahn | 17m |
KT Cloud describes rebuilding its public cloud into a “Next Architecture” that combines OpenStack with Kubernetes to present bare metal, VMs, and containers as… | ||
| Keynote: Open Source in the Age of AI | Jim Zemlin | 27m |
Open source is driving AI efficiency: open models like DeepSeek close the quality gap to closed systems while cutting training costs, frontier runs hit… | ||
| Enhancing Spatial Safety: Better Array-bounds Checking in C (and Linux) | Gustavo A. R. Silva | 01h 10m |
Shows concrete fixes: replaces __builtin_object_size mode-zero with mode-one to stop the original "bad vibes" memcpy overflow, adds annotations using… | ||
| Keynote: Jung-Woo Ha, Ph.D, Senior Secretary to the President, AI & Future Planning, AI Strategist | Jung-Woo Ha | 10m |
The keynote announces Korea's national AI capacity build-out: procurement of 260,000 NVIDIA GPUs and a program funding five companies to develop a KAI… | ||
| Upstream Kernel Hardening: Progress on Enabling -Wflex-array-member-not-at-end | Gustavo A. R. Silva | 35m |
GCC's -Wflex-array-member-not-at-end reveals that the kernel had ~60,000 warnings which collapse to about 650 unique issues; work reduces that to roughly 200… | ||
| Governing Open Source in the Enterprise: OSPO Strategies for Innovation and Compliance | David Hirsch | 28m |
Enterprise OSPO strategies that scale: OSPO adoption is up 32% year over year and Dynatrace documents 30–60,000 open-source components in its SBOM; governance… | ||
| Keynote: How Microsoft Builds and Runs on Open Source | Jay Lee | 06m |
Microsoft claims it is now a second-largest contributor to CNCF projects and the leading cloud vendor by CNCF contributions over the last three years, with… | ||
| Keynote: KOSSA: The Beginning of New Leadership | Hosung Shim | 06m |
KOSSA renames itself in 2025 and expands from software to include hardware, claiming the role of national connector across industry, defense and education; it… | ||
| Keynote: Owning Your Silicon Future Starts with Building on RISC-V | Divyang Agrawal | 10m |
Tenstorrent releases concrete RISC‑V hardware and software: an open-source Ocelot CPU (BOOM‑based) and a high‑performance Ascalon core on forthcoming developer… | ||
| Keynote: The Open Source Pathway: Democratizing AI's Benefits to Korea's Economy, Society, and Workforce | A. Hermansen | 06m |
Linux Foundation Research and a Meta partnership quantify concrete impacts: Korea gains $476 billion from AI-driven productivity, reports show, with >20%… | ||
| Keynote: Linus Torvalds, Creator of Linux & Git, in Conversation with Dirk Hohndel | Linus Torvalds, Dirk Hohndel | 30m |
Covers concrete claims about kernel evolution: maintenance is the ongoing work after 35-year project status, occasional removal of components (only a handful… | ||
| Building Enterprise Solutions in the AI-Native Era | Daniel Oh | 35m |
Red Hat AI claims an open source, hybrid-cloud platform that provisions sandboxes in minutes and runs models “wherever the data resides,” addressing cost and… | ||
| Securing Open Source With SBOM: From Visibility To Trust | Heejo Lee | 27m |
Claims an automated SBOM and vulnerability pipeline built as IoTcube.net (web UI plus back end called headbomb) that uses code‑level analysis algorithms… | ||
| The Open Source Imperative in AI: Transparency, Governance, and Global Collaboration | C. Xie, J. McCrosky | 29m |
Argues open source is necessary for auditability and sovereignty, introduces the TRUE framework to score models on Transparency, Reproducibility… | ||
| Vector Search Made Simple: Getting Started with OpenSearch for AI Applications | Dotan Horovits | 30m |
OpenSearch positions itself as a full vector database that handles trillions of vectors, offers both exact K-NN and ANN engines, and ships connectors for… | ||
| How To Talk To Your Lawyer about Open Source | Jimmy Ahlberg | 30m |
Ericsson reports open source rising from ~5% in 1995 to >90% of lines in some products and a fast increase in unique components introduced each year, so legal… | ||
| Accelerating Korea's Open Source Future: Workforce, Innovation, & AI | Anna Hermansen | 28m |
LF Research publishes three Korea-relevant reports: an APEC open source AI study, a state of sovereign AI survey, and a tech talent report with a sample size… | ||
| Panel Discussion - The Shift to Digital Money | D. Barbosa, J. Sung, L. Kim, A. Lim | 33m |
Discussion shows real milestones and open tools: LayerZero connects 150 blockchain networks and uses an OFT messaging standard; a recent bootstrap moved $550M… | ||
| Keynote: Rust in the Linux Kernel, Why? | Greg Kroah-Hartman | 14m |
Linux kernel adoption of Rust aims to eliminate classes of bugs by forcing error checks, resource lifetimes and locking at compile time rather than relying on… | ||
| Keynote: OpenSearch: From Project to Community-Driven Ecosystem—Building Sustainable Governance Together | A. Ross | 06m |
OpenSearch joins the Linux Foundation 13 months ago and moves from #70 to #19 in contributor rank, with a Technical Steering Committee of 15 members… | ||
| Code, Community, & Commerce: The Future of Payments with Solana’s Open-Source Ecosystem | B. Lee | 28m |
Solana presents an open-source stack delivering sub-second finality and about $0.0025 fees, highlighting the Firedancer C/C++ validator client to boost… | ||
| Policy Work Today and the New International Context | Ciarán O'Riordan | 34m |
Explains how regulators now treat software as infrastructure, using the Cyber Resilience Act (CRA) and AI Act as concrete pressures that change licensing into… | ||
| Copyleft and the EU CRA | Jimmy Ahlberg | 31m |
The EU Cyber Resilience Act imposes a legal duty to upstream any developed vulnerability fix for included software under Article 13‑6, with different outcomes… | ||
| HMSDK: Enhancing Software Support for CXL Memory | Honggyu Kim, Yunjeong Mun | 31m |
HMSDK provides two concrete kernel features now upstream: weighted interleaving (merged around Linux 6.9) and DAMON-based tiering (merged around 6.11)… | ||
| The Big-endian RISC-V Linux Adventure | Ben Dooks, Nazar Kazakov | 26m |
Project implements RISC‑V runtime big‑endian support end‑to‑end: added QEMU (≈120‑line patch) flags and page‑table handling, updated OpenSBI to set per‑hart… | ||
| Audit-Driven Signature Tracing for Smart Contract Safety | Jaehyung Lee | 33m |
Extracts reusable bytecode-level signatures from audit reports plus optional LLM or test-based summaries, publishes them as small JSON records (SPDX-like) with… | ||
| Orchestrating Real-Time Multimodal AI Agents with Rust | Miley Fu | 30m |
A Rust-based orchestrator (Echo server / Echo Kit) glues VAD, ASR (Whisper-compatible), LLMs, and TTS into a streaming pipeline that runs on an affordable… | ||
| A 10-Year Retrospective on Why Our Job Isn't Just About Compliance and Cost Savings | Seoyeon Lee | 15m |
A decade of OSPO work reframes the job as mediation not just compliance, documenting a 5‑year pattern of license lawsuits that rises then plateaus and noting a… | ||
| Who Ate My Resources? Trace LLM Performance in Real Time With OTel | Aditya Soni, Seema Saharan | 27m |
OpenTelemetry profiling plus an eBPF profiler traces a FastAPI + GPT2 demo end to end, showing a single forward call using ~39% CPU and one helper function… | ||
| Vibe Computing: AI-Powered Multi-Cloud Orchestration With Cloud-Barista | Son Seokho | 37m |
Cloud-Barista is an Apache‑2.0 Go platform that unifies heterogeneous cloud APIs via CB‑Spider and CB‑Tumble and adds an MCP server to map natural‑language… | ||
| Kernel CVEs are Alive, but Do Not Panic! | Greg Kroah-Hartman | 28m |
The kernel community becomes a CNA and now owns CVE assignment, issuing about 13 CVEs a day while fixing roughly 40 bugs daily; a two‑of‑three public review… | ||
| Verified Debian Packaging at Scale | Chris Arges | 25m |
Cloudflare proves every executable on 330+ city Debian hosts is signed and verified before execution by using Linux IMA appraisal and embedding signatures into… | ||
| Making RAG Chatbots Enterprise-Ready: Group Permissions and Pluggable Backends | J. Lee, S. Park | 24m |
Company-deployed RAG system Maru is open-sourced and already powers legal and ops workflows, using group-based access control mapped to OS-like folders, roles… | ||
| Driving Safety Forward: Lessons Learned From Deploying OSS in Real-world Automotive | Jaylin Yu | 34m |
EMQ ships an end-to-end open-source vehicle data loop that runs in mass production for ~10,000 customers, using MB-fuzzer and SP-fuzzer to find stateful MQTT… | ||
| Dependency Injection for On-Device AI Systems | Sangjung Woo, Abha Jaswal, Jinhwi Kim | 29m |
Architecture injects hardware, runtime and model details via JSON so apps run unchanged while backends swap under the hood. Implementation uses NNStreamer… | ||
| Your AI Agents Are Overprivileged: The Case for Fine-Grained Authorization | A. Jha, D. Ikehara | 30m |
Claim: agentic AI routinely gets broader data access than required and must enforce document-level permissions. Implementation uses OpenFGA relationship tuples… | ||
| Sokovan: An Acceleration-first Kubernetes-compatible AI Container Engine | Joongi Kim | 33m |
Sokovan is an acceleration‑first, Kubernetes‑compatible container engine that runs alongside K8s via a daemonset shim and custom operators, mapping logical… | ||
| RISC-V Bug Hunting | Ben Dooks | 36m |
Syzbot triage shows an occasional kernel OOPS where the SUM bit isn’t preserved across a context switch; investigation with QEMU, kernel disassembly and a… | ||
| Telco Supply Chain Security: Implementing ISO 18974 & SBOM | Haksung Jang | 31m |
SK Telecom lays out adopting ISO/IEC 18974 for programmatic open-source risk management, generating SBOMs in CI/CD using SPDX or CycloneDX, and running… | ||
| Detecting Double Free With BPF | Bojun Seo | 27m |
A BPF-based detector hooks malloc/free via BPF+uprobes, stores allocation records in BPF maps, and runs the trace logic in a separate process to keep the… | ||
| Smarter Code, Sneakier Risks: Supply Chain Security in the Age of AI | Lavakush Biyani | 26m |
AI coding tools like GitHub Copilot and CodeWhisperer speed development—developers go from 1–2 PRs to 5–7 PRs a day and whole codebases refactor in weeks—but… | ||
| Introduction and Consideration of Temporal Partitioning in Avionics With Open Source Eco-System | H. Kim, G. Kwon | 30m |
Defines concrete ARINC 653 requirements for avionics partitioning and shows gaps in current open source stacks, proposing a hypervisor-based architecture using… | ||
| DO-330 Qualification of Enhanced LLVM Structural Coverage Tool | Minji Park, Seojin Kim | 31m |
Boeing adapts an open-source LLVM structural coverage tool and pursues DO-330 qualification so coverage output counts toward DO-178C objectives; the enhanced… | ||
| Memory-Efficient LLM Inference on Edge Devices With NNTrainer | Eunju Yang, Donghak Park | 26m |
NNTrainer (github.com/nnstreamer/nntrainer) repurposes its device training schedulers to run huge LLMs by combining memory-efficient tensor scheduling with a… | ||
| Exploring Kepler's Next Chapter: Achieving Cloud Native Sustainability With MCP Integration | M. González, P. Rao | 28m |
Kepler runs as a DaemonSet reading CPU/DRAM energy from MSRs and /sys and attributes watts to pods via cgroups, Prometheus scrapes those metrics and Grafana… | ||
| nothing matches | ||