Amazon SageMaker Studio (web-based development)

Amazon SageMaker Studio is the purpose-built interface for data scientists and ML engineers. From the Tasks tab you can manage RayCluster, RayJob, RayCronJob, and RayService resources without writing Kubernetes manifests. Studio reads and writes the same KubeRay custom resources that the operator reconciles, so a resource you create in Studio is identical to one you apply with kubectl.

Studio suits teams that want data scientists productive on Ray without Kubernetes knowledge. Both surfaces act on the same resources, and you can move between them.

The following screenshot shows the Tasks tab with the Actions menu open on a Ray cluster.

Before you begin

Configure a SageMaker AI domain and grant it access to your cluster once, before anyone uses the Tasks tab. For more information, see Setting up Studio for Ray.