OpenMetadata

A unified metadata platform that lets you find, understand, and govern your data across every source. Think of it as the search engine for your data infrastructure: tables, dashboards, pipelines, ML models, all cataloged and lineage-tracked in one place. The core is Apache 2.0 and fully self-hostable.

Running it yourself means Docker Compose with a few services (Elasticsearch, MySQL/Postgres, Airflow for ingestion). Not trivial, but not painful either. 84+ connectors out of the box cover Snowflake, BigQuery, Redshift, dbt, Looker, and basically everything else. Data quality tests run no-code, and column-level lineage actually works across joins.

Solo devs won't need this. Small data teams (3-10 people) get massive value: finally knowing what tables exist, who owns them, and whether they're fresh. Enterprise teams get governance, classification, and access policies without buying Collibra or Atlan.

The catch: it's a platform, not a tool. You're adopting an entire metadata layer. If your data infra is two Postgres databases and a dbt project, this is overkill.