The last time I attended a conference was long before COVID-19, and never really thought much about attending conferences again until I saw the announcement for the 2024 Open Source Summit in Japan, which is organized and hosted by the Linux Foundation.
Conferences can be a tremendous source of inspiration because they allow you to meet new, smart, and fascinating people while also learning about their amazing stories and initiatives. Just made me question why I hadn’t attended a conference in such a long time.
Another reason for me to attend this conference is because Linux and OSS have always played an important role in my life, both professionally and personally. The ideology of OSS and Linux has always struck a chord with me, and the 2024 Open Source Summit Japan provided an excellent opportunity to learn firsthand about what is currently happening with Linux and OSS.
Attended talks
The conference hosted a large number of talks grouped by topics, or ‘summits’ as they organized them. I would have loved to join more of the other talks, however here is my schedule of the talks I attended over the course of the two days.
October 28 - Monday
| Time | Talk |
|---|---|
| 09:00 – 10:45 | Keynotes |
| 11:15 – 11:55 | Architecting Kubernetes-Based Internal Developer Platforms: Essential Patterns and Practices - Hiroshi Hayakawa, LY Corporation |
| 12:05 – 12:45 | Unlocking Agility with Open Source: A Guide for Modern Businesses - Younes Hairej, Aokumo Inc. |
| 14:00 – 14:40 | From Policy Enthusiast to Legislators’ Contact Point - Ciarán O’Riordan, OpenForum Europe |
| 14:50 – 15:30 | Securing the Future: Modernising Singapore Government IT Policy Through Open Source - Eugene Lim, Open Government Products |
| 15:40 – 16:20 | Unlocking Local LLMs with Quantization - Marc Sun, Hugging Face |
| 16:40 – 17:20 | A Practical Guide to Using International Standards for Open Source Procurement - Shane Coughlan, OpenChain Project |
| 17:30 – 18:10 | LFX Mentorship Showcase - Moderated by Shuah Khan, The Linux Foundation |
| 18:10 – 19:30 | Tux Trek Sponsored by Docker |
October 29 - Tuesday
| Time | Talk |
|---|---|
| 09:00 – 11:10 | Keynotes |
| 11:10 – 11:50 | Desktop Linux War Stories - Anna Aitchison, Callcare |
| 12:00 – 12:40 | Running Containers on a Resource Constrained Embedded Device - Jeff Shaw, Digi International |
| 14:00 – 14:40 | Data Contracts Are Good for AI - Jean-Georges Perrin, Bitol / AbeaData |
| 14:50 – 15:30 | Lessons Learned on Following Security Best Practices in Zephyr - Kate Stewart, The Linux Foundation |
| 15:50 – 16:30 | The Dark Side of AI: The Hidden Risks in Open-Source AI Models - Tal Folkman, Checkmarx |
| 16:40 – 17:20 | Helping the PyTorch Community Improve the Security Posture of this Popular AI Framework - Jeffrey Borek & Moriyoshi Ohara, IBM |
Talks that stood out / takeaways
Hot Topics at LF
- Central Bank Digital Currency
- Navigating Open Source Amid Geopolitical Challenges
- Understanding Open Source AI
- The confusion between security and trust
- New Research Studies
Private/Public Partnerships and Government’s role in OSS International bodies such as the United Nations and national governments are taking accelerated steps in their digital transformation, using FOSS principles and learnings to drive their effort. A great example is the Government of Singapore that uses OSS principles to define their IT security policies. For transparency, these policies are managed using Git where each agency can request changes by simply submitting a PR. Policy as a product, applying Git and DevOps to policy development.
OpenWallet
Automotive
SDV definition. Treat vehicle as server (or like a smartphone) in terms of capabilities, components and software delivery.
AI
Security Model and training data poisoning.
Compress huge LLM models using 8-bit quantization. Prune models by removing connections that do not improve the model. Train a smaller model based on the original model. Use 1 bit to store weights rather than the usual FP32 (floating point). 4x reduction and tests showed only a negligable degradation.
Current LLM’s are built on the English language. For adoption on other countries such as Japan, more research and study is required for non-English languages such as Japanese that has it’s own set of linguistic nuances. Lots of advocation for Open AI (not the company) and Open Source AI.
See you all next year at the 2025 OSS Japan in December!