Self-Play for LLMs, AI for Biology, Formal Verification, and More | YC Paper Club

It's hard to keep up with the latest AI research. That's why we started YC Paper Club — a small group of researchers, engineers, and founders who meet every two weeks at our Mountain View office to present and discuss new papers together. In this session, we cover whether scaling laws hold for protein biology, AlphaZero-style self-play for language models, streaming RAG for real-time voice agents, formal verification with Lean, and why one founder thinks programming with agents is exactly like playing a real-time strategy game. Stay tuned for more. Interested in joining a future Paper Club? Apply here: https://events.ycombinator.com/ycpaperclub Apply to Y Combinator: https://www.ycombinator.com/apply Work at a startup: https://www.ycombinator.com/jobs 00:00 — Introduction by Francois Chaubard 05:47 — Yasa Baig: A World Model of Protein Biology (https://biohub.ai/esm/protein/about) 25:38 — Luke Bailey: Scaling Self-Play with Self-Guidance (https://arxiv.org/pdf/2604.20209) 37:51 — Arnab Maiti: Stream RAG: Instant and Accurate Spoken Dialogue Systems with Streaming Tool Usage (https://arxiv.org/pdf/2510.02044) 47:40 — Robert George: Lean for Science: How Formal Proofs Can Change Mathematics, AI, and Scientific Computing (https://arxiv.org/abs/2602.22631) 58:52 — Lukens Orthwein: Founder AI Hacks: Programming is an RTS Game Now 1:16:07 — Closing Remarks

Francois ChaubardhostYasa BaigguestArnab MaitiguestRobert GeorgeguestLukens Orthweinguest

Jun 12, 20261h 16mWatch on YouTube ↗

EPISODE INFO

Released: June 12, 2026
Duration: 1h 16m
Channel: Y Combinator
Watch on YouTube: ▶ Open ↗

EPISODE DESCRIPTION

It's hard to keep up with the latest AI research. That's why we started YC Paper Club — a small group of researchers, engineers, and founders who meet every two weeks at our Mountain View office to present and discuss new papers together. In this session, we cover whether scaling laws hold for protein biology, AlphaZero-style self-play for language models, streaming RAG for real-time voice agents, formal verification with Lean, and why one founder thinks programming with agents is exactly like playing a real-time strategy game. Stay tuned for more. Interested in joining a future Paper Club? Apply here: https://events.ycombinator.com/ycpaperclub Apply to Y Combinator: https://www.ycombinator.com/apply Work at a startup: https://www.ycombinator.com/jobs 00:00 — Introduction by Francois Chaubard 05:47 — Yasa Baig: A World Model of Protein Biology (https://biohub.ai/esm/protein/about) 25:38 — Luke Bailey: Scaling Self-Play with Self-Guidance (https://arxiv.org/pdf/2604.20209) 37:51 — Arnab Maiti: Stream RAG: Instant and Accurate Spoken Dialogue Systems with Streaming Tool Usage (https://arxiv.org/pdf/2510.02044) 47:40 — Robert George: Lean for Science: How Formal Proofs Can Change Mathematics, AI, and Scientific Computing (https://arxiv.org/abs/2602.22631) 58:52 — Lukens Orthwein: Founder AI Hacks: Programming is an RTS Game Now 1:16:07 — Closing Remarks

SPEAKERS

Francois Chaubard
host
YC Paper Club facilitator at Y Combinator who moderates the session and introduces presenters.
Yasa Baig
guest
AI-for-biology researcher presenting on applying machine learning to drug discovery and related problems.
Arnab Maiti
guest
Presenter covering a Meta paper on streaming retrieval-augmented generation (RAG) for low-latency/voice AI settings.
Robert George
guest
Presenter discussing formal verification and the Lean proof assistant for verified intelligence.
Lukens Orthwein
guest
Channel AI founder discussing agentic programming and self-play-inspired training for coding agents.

EPISODE SUMMARY

In this episode of Y Combinator, featuring Francois Chaubard and Yasa Baig, Self-Play for LLMs, AI for Biology, Formal Verification, and More | YC Paper Club explores yC Paper Club explores biology AI, self-play, RAG, verification, hacks Protein language modeling is shown to follow scaling-law-like behavior when training data is massively expanded (e.g., metagenomics), enabling strong structure/function signals from sequence-only pretraining and even interpretable biological features.

RELATED EPISODES