Beta Release of the Fusion Data Platform: A Community Resource for Fusion Data Science

The Fusion Data Platform (FDP) is a federated data infrastructure designed to simplify access to curated experimental and simulation datasets and support reproducible AI/ML workflows for fusion energy research. FDP’s primary goal is to enable seamless discovery, access, and use of large-scale fusion data from distributed sources, while ensuring comprehensive provenance tracking.

Following last year’s alpha release, the current beta release offers expanded federated data capabilities and new workflow tools, and is now available as a community-wide resource, with new users invited to use the platform. Built on the Open Science Data Federation, FDP provides secure access to petabyte-scale datasets from multiple tokamaks, including DIII-D and MAST. The platform supports mapping device data to the IMAS data schema, facilitating multi-device analysis.

FDP has been demonstrated at multiple HPC centers, including SDSC and NERSC, with support for single-command installation into user workflows. The beta release introduces simplified user interfaces and integration with HPC job schedulers, making it easy to integrate with existing workflows. Future developments will expand support for additional fusion facilities and include AI agent integration for streamlined data processing. This poster highlights FDP’s capabilities and example workflows, including turbulence surrogate modeling and safe operational regime identification, and invites new users to explore and contribute to the platform.