hora

Why we included this project

Hora is a Rust library for approximate nearest neighbor search, useful when you have to match millions of vectors quickly. It offers several index types, including HNSW, SSG, and product-quantization inverted files, so you can trade speed against recall depending on your data. The distance calculations are SIMD-accelerated, which keeps performance close to what you would expect from C++. Python, JavaScript, and Java bindings mean you can use it without rewriting an existing application, and WebAssembly support lets the same code run in the browser. If you prefer to choose and tune the index algorithm yourself rather than rely on a managed vector database, this is a practical building block.