Adding missing index types to LanceDB's Python and Node SDKs

Project Showcase
1

Hi all, sharing some open-source work from the last few weeks in case it’s useful to anyone here working with vector search or AI infra.

LanceDB is an embedded vector database with a Rust core and Python/Node SDKs. The core supported several index types that weren’t exposed in the SDKs, so things you could do in Rust weren’t reachable from Python or JS. I’ve been closing that gap:

  • CombinedFieldsQuery for full-text search: merged
  • ZoneMap scalar index in the Python API: merged
  • BloomFilter scalar index (Python) and NGram index (Node.js): open, in review

I’ve also landed smaller fixes in supabase-py (realtime channel rejoin backoff), a2a-python, and pypa/distlib, and reported a bug that got fixed in FastMCP.

Mostly I wanted to share that the SDK layer of a big project is a very approachable on-ramp. The Rust core already had the logic, so the work was exposing it cleanly with tests and docs. Happy to talk through how I spotted these gaps if anyone’s looking for a way into a larger codebase.

If you use LanceDB from Python or Node and want a particular index or query type surfaced, tell me. It’s a good source of well-scoped PRs.

My PRs and profile: BHAWESHBHASKAR · GitHub