Huge genomic datasets are being created all around the world, and their scale is accelerating. But these data gain greater meaning when analyzed in concert with other datasets stored in institutions around the world. Due to data residency restrictions, regulatory barriers, and sheer data volume, it is impossible to effectively centralize all of these data in one place. In order to achieve regional and global use of many data sets in concert, we must overcome these challenges with a new approach to managing, analyzing and sharing sequencing data: Federated Computing. Federated Computing is difficult from a technical perspective because of the variety of IT infrastructures and workflow engines available, which makes reproducibility across environments nearly impossible, and from a practical perspective because of privacy and competitive concerns among researchers. Federated Computing becomes easier with a scalable, open source, multi-platform, standards-based biomedical big data computing platform that can be deployed in public cloud, private cloud, and HPC environments, and enables bit-for-bit reproducibility of analyses across every deployment. We present Arvados (http://arvados.org), a free and open source platform for managing and processing biomedical data designed for scale, reproducibility, and federation. Workflows and queries can travel across multiple Arvados clusters, running exactly the same way on each one, regardless of the underlying compute & storage infrastructure.