rah-qdrant-integration

A community add-on that brings Qdrant vector search to RA-H OS by Bradley Morris.

This project extends RA-H OS with a Qdrant vector database backend, enabling high-performance semantic search alongside the existing SQLite knowledge graph. It was built as a practical solution for environments where sqlite-vec doesn't work reliably — specifically ARM64 devices (like the NVIDIA DGX Spark) and NFS-mounted storage.

This is not a fork or replacement. RA-H OS remains the core platform. This integration runs alongside it, adding vector search capabilities while preserving the full node/edge/dimension graph that makes RA-H OS powerful.

• [RA-H OS](https://github.com/bradwmorris/ra-h_os) by Bradley Morris — the knowledge graph platform this integration extends. All graph architecture, node/edge schema, skills system, and core design are Bradley's work.
• [Qdrant](https://qdrant.tech/) — open-source vector database used for semantic search.
• Integration layer by Nathan Maine — the Qdrant adapter, Docker configuration, and search utilities in this repo.

RA-H OS uses SQLite for its knowledge graph, which is excellent for graph queries and relationship traversal. However, for large-scale semantic search (50K+ chunks), a dedicated vector database provides:

• ARM64 native support — runs on NVIDIA DGX Spark, Raspberry Pi, Apple Silicon without compatibility issues
• Docker-based — portable across any environment, no native compilation needed
• NFS/network storage compatible — works reliably over mounted network drives
• Horizontal scaling — handles millions of vectors without degradation
• Payload filtering — filter search results by metadata (creator, source, tags) at query time

┌─────────────────────────────────