Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn about the latest advancements in observability for Ray distributed AI workloads through this 26-minute conference talk from Ray Summit 2025. Discover why contextual observability is essential as Ray becomes the standard framework for distributed AI applications, particularly when dealing with the inherent complexity of large, multi-node systems. Explore how purpose-built observability tools help diagnose critical issues including resource bottlenecks, task failures, and memory pressure. Examine major new improvements to observability on Anyscale, featuring scalable and persistent dashboard views for Ray Core, Ray Train, and Ray Data, along with the underlying architecture that powers these dashboards while maintaining security and keeping all data within your cloud environment. Get introduced to the open-source Ray Export API that enables you to persist, analyze, and integrate Ray dashboard events into your own monitoring and analytics systems. Watch a live demonstration showing how to debug real-world issues ranging from out-of-memory errors to inefficient resource utilization, making Ray workloads more transparent, reliable, and easier to optimize than ever before.
