FerresDB – High-performance vector search engine built in Rust

Summary: FerresDB is a persistent vector search engine designed for semantic search and retrieval-augmented generation (RAG). It uses HNSW for indexing, supports gRPC/REST APIs, WebSockets, and includes crash recovery via WAL.

What it does

FerresDB indexes vectors using HNSW and provides access through APIs and a built-in Dashboard, ensuring fast and reliable semantic search with crash recovery.

Who it's for

It targets developers needing a high-performance vector search engine with observability and API support for semantic search applications.

Why it matters

It solves the need for a fast, reliable vector search engine with developer-friendly features and crash recovery mechanisms.