Manage Amazon SageMaker HyperPod Spaces directly from SageMaker Studio
aws ·
AWS has introduced a new capability allowing users to manage Amazon SageMaker HyperPod Spaces directly from the SageMaker Studio user interface. This update eliminates the previous reliance on command-line tools like the HyperPod CLI or kubectl, enabling data scientists and ML engineers to launch development environments with just a few clicks. Users can now create, configure, start, stop, and open JupyterLab or Code Editor spaces through a guided form within the IDE and Notebooks tab. The integration supports various features including fractional GPU allocations, persistent storage via EBS, and remote connections to local VS Code instances via SSH-over-SSM. Administrators can further customize the experience by enabling optional capabilities such as Task Governance for compute quotas and Karpenter over-provisioning to reduce startup latency. These enhancements bridge the gap between high-performance infrastructure and interactive model development, fostering greater efficiency and self-service access for ML teams.
aws