endee

Why we included this project

Endee is a self-contained vector database built for retrieval-heavy production workloads, so it is worth a look if you are choosing the storage layer for a RAG pipeline or a semantic search service. It stores high-dimensional embeddings, exposes an HTTP API for indexing and querying, and layers metadata filtering and sparse retrieval on top of dense vector search, so hybrid relevance does not require stitching several services together. The C++ core is tuned for specific CPU targets such as AVX2, AVX512, and NEON, which matters for teams running on bare metal or Docker who care about throughput per node rather than a managed SaaS. It also ships documented backup flows and runtime logging, the operational details you tend to need before moving a search service into production. Teams experimenting with agent memory or retrieval augmented generation will find it a compact, self-hostable option worth profiling against their own data.