Loom, the data registry, repository, and workbench for Hadoop, is the product of our deep experience designing, building and managing data fabrics for the world's largest organizations and most complex data universes. We empower data scientists to answer the questions of today and tomorrow. A browser-based application, Loom provides data scientists with a workbench user interface for working with disparate Hadoop-based datasets. Loom brings order to the chaos of managing Hadoop-based data sets, recording workflows, tracking data lineage, and capturing provenance metadata. Loom converts SQL into MapReduce jobs using an extensible workflow paradigm, automatically updating the Loom Registry with workflow results.