Faiss

FAISS is Meta's vector similarity search library built to handle massive scale, we're talking billions of vectors with GPU acceleration. If you're building semantic search, RAG pipelines, or anything that needs to find similar embeddings fast, this is the industry standard solution. The 31K+ GitHub stars aren't hype, it's legitimately faster than most alternatives for pure vector similarity at scale. The tradeoff is it's focused purely on vectors, so if you need complex metadata filtering alongside similarity search, you might want a vector database instead. This skill gives you the template for integrating FAISS into Claude projects when raw performance on large embedding datasets matters more than bells and whistles.