Developed Zarr v2 support for Ray Data in the ray-project/ray repository, introducing the ray.data.read_zarr() API and implementing ZarrV2Datasource to enable reading Zarr v2 stores and accessing chunk metadata through the standard Ray Data API. The work focused on expanding data source compatibility and streamlining ingestion of large scientific datasets, thereby reducing data preparation time and improving pipeline reliability. Leveraged Python for backend integration, API development, and comprehensive unit testing to ensure robust functionality. Collaborated across teams to update documentation and provide usage examples, supporting scalable, metadata-rich data processing workflows within the Ray Data ecosystem. No bugs were reported.