Adapting Ray to Scale RL Training for Nemotron | NVIDIA | Ray Summit 2026

Nemotron is one of the leading open model families pushing the frontier of large-scale reinforcement learning. Training it with RL means coordinating complex datasets, long sequence lengths, a 550B-parameter architecture, and workloads spanning thousands of GPUs. At Ray Summit 2026, Wenwen Gao and Saurabh Mishra from NVIDIA share how they evolved Ray into a critical orchestration layer for RL at this scale, walking through key improvements across scheduling, fault tolerance, workload coordination, and large-scale training operations that made massive RL jobs more reliable and efficient. You'll leave with practical best practices and hard-earned lessons for adapting Ray-based infrastructure to large-scale RL training workloads. Liked this video? Check out other Ray Summit breakout session recordings: https://www.youtube.com/playlist?list=PLZNaJyYZRPdo Subscribe to our YouTube channel to stay up-to-date on the future of AI: https://www.youtube.com/c/anyscale 🔗 Connect with us: LinkedIn: https://www.linkedin.com/company/joinanyscale/ X: https://x.com/anyscalecompute Website: https://www.anyscale.com/