Thank you to everyone who joined our LF Edge InfiniEdge AI session during Open Source AI Week 2025! This fast paced workshop explored how to build Generative AI on distributed edge clouds and drew more than 180 participants to 3120 Scott Blvd in Santa Clara, CA.

During the session, speakers from LF Edge, LF AI & Data and the broader open source community shared hands on lessons on running AI closer to users. We discussed edge inference, data pipelines, privacy enhancing techniques and AI agents that run on devices. Attendees included AI/ML engineers, cloud practitioners, OSS contributors, maintainers and product leaders exploring real world edge GenAI use cases.

What you’ll learn:

• • How to run GenAI workloads on geo distributed edge clouds

• • Design patterns for privacy, latency and reliability

• • Practical demos with InfiniEdge AI and related open-source software

Agenda highlights:

• • Welcome & OS AI Week context (5 m)

• • InfiniEdge AI updates & roadmap – Tina Tsou, TSC Chair, InfiniEdge AI (LF Edge) (10 m)

• • Deploy SGLang on NVIDIA DGX Spark – Pen C Li (NVIDIA) & Jerry Zhou (SGLang) (10 m)

• • Scaling AI infrastructure & edge considerations – Ashok Prakash, Oracle (20 m)

• • Provable scaling laws of feature emergence from learning dynamics of Grokking – Yuandong Tian, former Research Scientist Director, Meta (15 m)

• • GenAI value proposition in advertising & marketing – Rick Cao, Head of Ads GenAI, Snap (10 m)

• • Hands on experience of SGLang on NVIDIA GPUs – Xinyuan Tong, SGLang (40 m)

• • Q&A and contributor pathways (15–20 m)

If you missed the live workshop or want to revisit the material, you can watch the full recording and explore the resources below:

• • Zoom recording

• • Google Drive folder with slides/resources

Thank you for being part of the InfiniEdge AI community and helping advance open source AI at the edge!

Acknowledgments

A huge thank you to NVIDIA for powering the DGX Spark demonstrations and to the SGLang community for showcasing how open source innovation accelerates AI at the edge. Your collaboration and contributions made this InfiniEdge AI session possible and inspiring for all participants.