Mandy Chessell, IBM | Dataworks Summit EU 2018
summary
Overview of an enterprise metadata stack: a graph-backed catalog built on Apache Atlas on JanusGraph that stores dataset lineage, classifications and governance rules, plus open metadata and governance libraries that let vendors and runtimes push and pull metadata. Atlas handles ML artifacts (models, training data, drift checks), supports HDFS or other storage, and the team defines a metadata exchange protocol to federate catalogs for GDPR audits, explicable-AI tracing, and multi-cloud deployment.
AI-written summary · about
