[GitHub Trending] RyanCodrai/turbovec
Scored daily by a customisable AI persona to surface the most relevant engineering leadership news.
Rust-based vector index with Python bindings is technically deep and relevant to data engineering.
turbovec is a Rust vector index with Python bindings that compresses a 10M-document corpus from 31 GB (float32) to 4 GB using Google Research's TurboQuant algorithm, a data-oblivious quantizer matching the Shannon lower bound without training. It outperforms FAISS IndexPQFastScan by 12–20% on ARM via hand-written NEON/AVX-512BW kernels, supports online ingest with no rebuilds, and enables filtered search within SIMD kernels using id allowlists or bitmasks. Drop-in integrations for LangChain, LlamaIndex, Haystack, and Agno let teams swap in turbovec without pipeline changes.
RyanCodrai