ParadeDB
An open-source Postgres search extension for full-text and vector retrieval, filters and aggregations. Run it locally or self-host it under AGPL-3.0.
ParadeDB adds full-text, vector and hybrid search to the Postgres database that holds your application data. It's for developers who need search and analytics without maintaining a separate Elasticsearch cluster or synchronizing a second copy of their data. It runs as the pg_search extension in self-managed Postgres, including local Docker deployments.
Search uses standard SQL, so you can combine ranked results with joins, filters and existing database logic. Full-text capabilities include tunable BM25 scoring, typo tolerance, highlighting, phrase matching, fuzzy search, proximity queries and regex queries. Vector retrieval and hybrid search support searches that combine lexical and semantic relevance. Facets and aggregations let applications group results and calculate metrics alongside retrieval.
The index combines inverted, vector and columnar structures within Postgres. Rust libraries Tantivy and Apache DataFusion provide search and analytical processing. Application data stays in your database, with ACID transactions and read-after-write guarantees; search doesn't require a sidecar process, an external cluster or schema changes.
ParadeDB is open source under AGPL-3.0, with community and enterprise self-managed offerings. It works with Drizzle, Django, SQLAlchemy, Rails and EF Core. For AI development workflows, it also provides agent skills, an MCP integration and a Cursor plugin.
