Scaling AI & Machine Learning Workloads With Ray on AWS, Kubernetes, & BERT
summary
Open source Ray on AWS and Kubernetes shows concrete configs and tools: Ray AIR for training and serving, Ray Tune for hyperparameter search, Ray Serve for model DAGs, and KubeRay on EKS with Terraform blueprints. Demonstrations include Ray up / Ray down automation, Ray cluster dump diagnostics, conda-as-a-service environments, and a claim of eight GPUs at 100% for 4 hours after a single “turbo” flag tweak, plus notes on disk spill behavior and Spark vs Ray performance.
AI-written summary · about
chapters (6)
- 0:00Introduction and poll
- 3:54What is Ray and its 2.0 release
- 11:41Ray’s role in GPU utilization
- 17:23Ray cluster management on AWS
- 24:04Ray Serve and future directions
- 27:14Summary and resources
related talks
28mRay on Public-Cloud k8s - experiments, lessons & best practices
22mRay, a Unified Distributed Framework for the Modern AI Stack
26mIntroduction to Distributed ML Workloads with Ray on Kubernetes
31mNext-Generation Data Science Workflows Using Ray
44mPanel: ML Platform on Ray
28mScaling Multimodal Data and Reinforcement Learning with Ray
31mHosting AI models with Ray and GKE
35mOperating a Production-Scale Physical AI Data Platform with Ray