Introducing MDMongoose — MDM without Madness
Master data management usually means another platform next to your lakehouse: another database, another copy of your data and another set of permissions to keep in sync. MDMongoose takes the opposite route — it lives inside Databricks.
Why another MDM?
Most of the data that master data is built from already sits in the lakehouse: ERP and CRM extracts, HR feeds, product files. So we asked a simple question: what if the master data stayed there too?
MDMongoose keeps the whole hierarchy — reference data, loaded records, golden records and even its own definitions — in one Delta table. A web app on Databricks Apps lets data stewards browse and edit it, Spark Declarative Pipelines load source changes incrementally and publish materialized views, and serverless jobs build the golden records. One bundle deploys everything, and it runs on Databricks Free Edition.
What you can do today
- Browse and edit one hierarchy of everything, with typed properties and structure rules.
- Edit hundreds of nodes at once in the Excel-like Edit Children grid.
- Load source tables incrementally from their change data feed, with data-quality expectations.
- Build golden records with matching and per-property survivorship rules — and a dry run that explains every value.
- Export to CSV and Excel, or consume typed materialized views in Unity Catalog.
- Control access with Unity Catalog groups.
What comes next
On this page we will publish release notes and articles about master data management on Databricks — hierarchy design, matching strategies and running MDM at scale. Start with the documentation, and tell us what you would like to read about.