AWS expands HyperPod support for Ray AI workloads

AWS says SageMaker HyperPod now adds built-in observability, resilient training, accelerated inference and managed development environments for Ray workloads on Amazon EKS. Users can manage Ray clusters in SageMaker Studio, connect interactive development environments, and use provisioned Grafana dashboards and secure Ray Dashboard links. AWS also says node recovery, hung-job detection, tiered checkpointing and task governance are available for training, while Ray Serve can use tiered KV cache and deploy JumpStart models. Existing open-source Ray code runs unchanged.