ParadeDB

Developers no longer have to face the dilemma of choosing between Postgres and a separate search engine. Mithun Chicklore Yogendra from ParadeDB unveils **pg_search**, an extension bringing advanced BM25 full-text search capabilities directly into PostgreSQL. #DeveloperTools #SearchEngine #Postgres

We're launching Agent Skills and MCP Server for ParadeDB. These will empower your AI coding agent to produce top-notch, accurate ParadeDB code effortlessly. For Skill, go to: https://lnkd.in/dPyGTTQT For MCP:

Give me production search, but it's just Postgres (Supabase edition 👀) https://lnkd.in/e_WhFqTU

Ankit mittal (ex-Instacart) walks through how to build personalized search entirely inside PostgreSQL. The example is simple: a movie search for “king” could mean LOTR: Return of the King or The King’s Speech. Which one you want should depend on the user’s genre preferences. The implementation uses a straightforward two-step process: 1. Retrieve: use BM25 to quickly pull a small set of strong lexical matches. 2. Rerank: reorder those results using user embeddings and cosine similarity. Both steps run in SQL, in Postgres. No external services. No data movement. https://lnkd.in/eFccYynk

As part of our v0.20.0 release late last year we did a lot of work on increasing write throughput. Adding any index to a Postgres table trades off write speed for read speed, and BM25 indexes aren't any exception. But because they map to immutable segments the impact can actually be a lot more than traditional indexes like Btree. So we set out to make things better .... https://lnkd.in/e5uiyVDN

V2API goodness!

Faceting is one of those features that looks simple on the surface (it's just counts next to search results, right?), but is surprisingly hard to make both easy for developers to use, and efficient under the hood. We just shipped native faceting for ParadeDB. It runs inside Postgres, in the same pass as your search query, and it’s an order of magnitude faster than doing it manually in SQL over a large result set. If you care about search performance (or just love reading about making Postgres do new things), give it a read 👇 https://lnkd.in/e6K6HMSk

We've teased our V2 SQL API a bit, but here's a full post looking at everything you can do with it. Check out the side by side of the CREATE TABLE for old and new APIs, it's so much better ❤️. https://lnkd.in/gVBqamrq

We're excited to announce the release of v0.20.0. This release brings: - Official support for accelerated facets in ParadeDB - An order of magnitude improvement to our write throughput - A brand new SQL interface that's more SQL-native and ORM-friendly https://lnkd.in/eQ7jQbEk

Most benchmarks claiming Tantivy is faster at search than Lucene come from people building Tantivy (like Paul Masurel), or building on Tantivy (like us). Here’s a nice change: Adrien Grand (a Lucene committer) explains exactly why Tantivy pulled ahead, and what Lucene is learning from it. Great read for anyone building or tuning search systems: https://lnkd.in/eshenrg2