[Opening Keynote] State of Ray
summary
Ray reports concrete growth and engineering wins: 480 contributors and production wins like XGBoost running 5x faster on Ray and Dask workloads running 13x faster while clusters scale 7x and job cost falls to 30% of original. Scalability work supports petabytes/hour throughput and deployments across 200,000 cores. New runtime features include Ray Client for remote development, environments for consistent dependency syncing, object spilling and memory improvements, and a revamped serving system (rayserve) for production.
AI-written summary · about
related talks
15mRay: A Distributed Compute Engine for AI
22mRay, a Unified Distributed Framework for the Modern AI Stack
28mScaling AI & Machine Learning Workloads With Ray on AWS, Kubernetes, & BERT
16mKeynote: An Open-Source Stack for AI Compute: Kubernetes + Ray + PyTorch + vLLM
29mBuilding Scalable AI Agents with Ray
28mServing LLMs for Agents at Scale with Ray + vLLM
34mScaling RL with Ray, Kubernetes, and TPUs
17mFrom Agents to Action: Scaling Agentic AI with Ray on Intel Xeon