Building Open AI Agents: Inference, Memory & Coding Systems
On October 23 in San Francisco, join us for an evening of lightning talks, food, and open discussion as part of Open Source AI Week 2025.
This edition of our community series will focus on making LLMs efficient in practice and designing agents that scale.
📆 Agenda
6:00–6:20pm: Arrival, food & networking
6:20–6:40pm: Alex Shaw (Laude Institute): Terminal-Bench: How Good Are CLI Agents?
6:40–7:00pm: Seiji Eicher (Anyscale): Efficient Multi-Node Orchestration for Sparse MoE Model Serving
7:00–7:20pm: Smit Kadvani (Meta): Throughput Efficiency Using Disaggregated Inference
7:20–7:40pm: Kerim Kaya (Dria) : Evolutionary Coding Agents
7:40–8:00pm: Zhuohao Li (SGLang): Optimizing Large Scale RLHF
Over the past year, our events in NYC and SF have brought together 1,500+ developers and researchers, with speakers from Google Labs, Nvidia, Modular, Meta, Snowflake, AI2, Warp, and more.
This time, we’re gathering during Open Source AI Week to keep pushing the frontier of open and practical AI.
Whether you’re building infrastructure, fine-tuning models, or experimenting with new agent designs, this is the place to connect with the people shaping the future of AI.