HelixDB: The Rust-Built Graph-Vector Database Unifying AI Memory, Knowledge Graphs, and RAG

HelixDB is an open-source graph-vector database built in Rust for AI memory, knowledge graphs and RAG. Instead of keeping embeddings in one database and relationships in another, Helix stores graph and vector data in the same engine, while also supporting KV, documents and relational data. Queries can combine graph traversal, filtering and vector search in one request, and the current SDKs for Rust, TypeScript, Go and Python send the same structured query AST directly to the server. It can run in memory, on disk, or against S3-compatible object storage.
5.7k stars · Apache 2.0
GitHub Repo: https://lnkd.in/dvqQ-zFK

HelixDB is confirmed to be an open-source graph-vector database built in Rust, designed for AI memory, knowledge graphs, and RAG applications. It combines graph and vector data models natively, while also supporting KV, document, and relational data — verified via the official repo and independent write-ups. Queries are authored through SDKs for Rust, TypeScript, Go, and Python, which all send a structured query (JSON AST) to the server — confirmed by the official README and docs. It supports in-memory and on-disk local instances, with Cloud deployments scaling across object storage (S3-compatible) — confirmed.