Uploaded November 2025 | Updated September 2026, 2 weeks ago
Slides: drive.google.com/file/d/1mUygPmRBv4N44i6hq6lDm6eZvxJ52T1a/view?usp=sharing
At Ray Summit 2025, Ed Oakes and Jiajun Yao from Anyscale share the major advancements made over the past year to enhance the performance, resiliency, and observability of Ray for cutting-edge, large-scale applications.
They walk through key improvements across the Ray Core runtime, including:
Ray Direct Transport, a next-generation communication layer that dramatically optimizes data transfer between accelerators
Native resource isolation with cgroups, enabling stronger workload separation and more predictable performance
Greater resiliency to network failures, ensuring long-running distributed jobs remain stable in real-world environments
Significant observability upgrades, providing deeper insight into system behavior at cluster scale
Ed and Jiajun also offer a first look at the Ray Core 2026 roadmap, outlining the next wave of features and architectural enhancements designed to support increasingly complex, performance-critical distributed workloads.
Attendees will gain a clear understanding of how Ray is evolving to meet the demands of next-generation AI systems and what to expect from the platform in the year ahead.
Liked this video? Check out other Ray Summit breakout session recordings youtube.com/playlist?list=PLzTswPQNepXllnU0C36WtkC0dqkAoDulh
Subscribe to our YouTube channel to stay up-to-date on the future of AI! youtube.com/c/anyscale
đź”— Connect with us:
LinkedIn: linkedin.com/company/joinanyscale
X: https://x.com/anyscalecompute
Website: anyscale.com
Slides: drive.google.com/file/d/1mUygPmRBv4N44i6hq6lDm6eZvxJ52T1a/view?usp=sharing
At Ray Summit 2025, Ed Oakes and Jiajun Yao from Anyscale share the major advancements made over the past year to enhance the performance, resiliency, and observability of Ray for cutting-edge, large-scale applications.
They walk through key improvements across the Ray Core runtime, including:
Ray Direct Transport, a next-generation communication layer that dramatically optimizes data transfer between accelerators
Native resource isolation with cgroups, enabling stronger workload separation and more predictable performance
Greater resiliency to network failures, ensuring long-running distributed jobs remain stable in real-world environments
Significant observability upgrades, providing deeper insight into system behavior at cluster scale
Ed and Jiajun also offer a first look at the Ray Core 2026 roadmap, outlining the next wave of features and architectural enhancements designed to support increasingly complex, performance-critical distributed workloads.
Attendees will gain a clear understanding of how Ray is evolving to meet the demands of next-generation AI systems and what to expect from the platform in the year ahead.
Liked this video? Check out other Ray Summit breakout session recordings youtube.com/playlist?list=PLzTswPQNepXllnU0C36WtkC0dqkAoDulh
Subscribe to our YouTube channel to stay up-to-date on the future of AI! youtube.com/c/anyscale
đź”— Connect with us:
LinkedIn: linkedin.com/company/joinanyscale
X: https://x.com/anyscalecompute
Website: anyscale.com
