github.com/amundsen-io/amundsen ↗
Amundsen is a metadata driven application for improving the productivity of data analysts, data scientists and engineers when interacting with data.
Open this visualization on its own page →
Contributors
224
Lines of Code
14,612
From
2019-02-04
To
2023-03-06
About amundsen-io/amundsen
Amundsen is a metadata-driven platform designed to help data analysts, data scientists, and engineers discover and understand data resources. It indexes tables, dashboards, streams, and other data entities, then powers a search interface that ranks results based on usage patterns—similar to how Google surfaces frequently accessed content first. The system is named after explorer Roald Amundsen and is hosted by the LF AI & Data Foundation as an open-source project.
The platform consists of multiple interconnected services: a Flask and React frontend for user interaction, an Elasticsearch-backed search service for querying metadata, a metadata service leveraging Neo4j or Apache Atlas for storage, and a databuilder library for ingesting and indexing metadata. Users can ingest data either through Python scripts or Apache Airflow DAGs. Amundsen supports a comprehensive range of data sources including major databases like Snowflake, BigQuery, Redshift, and PostgreSQL, as well as data orchestration platforms like Airflow and visualization tools like Tableau and Superset. The platform tracks tables, dashboards, ML features, and people from HR systems, with metadata enriched through features like column statistics and data previews.
Amundsen has been adopted by dozens of major organizations including Lyft, Square, Instacart, Snap, and Netflix. The project maintains an active community with monthly meetings and extensive documentation, offering multiple backend storage options and integration patterns to suit different enterprise architectures.