weaviate

Why we included this project

Most teams that build retrieval-augmented generation, semantic search, or recommendation features eventually need a place to store and query embeddings, and Weaviate is a strong open-source choice for that. It keeps both the original objects and their vectors, so similarity search, structured filters, keyword matching, and reranking can all happen in one query instead of across separate systems. You can either let Weaviate vectorize data at import through integrated model providers or bring your own pre-computed embeddings, which gives you flexibility in how you feed it. Multi-tenancy, replication, and role-based access control are built in, and it deploys on Docker or Kubernetes, so it holds up in production rather than just in a notebook. If you want a self-hosted vector store that can grow with a real application, this is worth a close look.