KeraDB is a single-file, embedded document database designed for simplicity and performance. Think SQLite, but for JSON documents with built-in vector search capabilities!
| Feature | Description |
|---|---|
| Single-file database | One .ndb file contains everything |
| Fast | Written in Rust with zero-cost abstractions |
| Memory-safe | Rust's guarantees prevent crashes and data corruption |
| Vector Search | HNSW index for fast approximate nearest neighbor search |
| Delta Compression | LEANN-style compression with up to 97% storage savings |
| Multi-language SDKs | Rust, Node.js, Python, Go, C#, and more |
| Repository | Description |
|---|---|
| keradb | Core database engine and CLI |
| keradb-labs | Desktop GUI app with Tauri + React |
| keradb.github.io | Documentation website |
# Install KeraDB CLI
curl -sSf https://keradb.github.io/quickstart.sh | sh
# Create a database and start the shell
keradb shell myapp.ndb
# Insert and query documents
keradb> insert users {"name": "Alice", "age": 30}
keradb> find users- Sub-40us vector search on 10K vectors
- ~50K ops/sec document lookups
- ~9K vectors/sec bulk insert rate
- Up to 97% compression for similar vectors
Check out our latest benchmark here
We welcome contributions! Check out our repositories and feel free to open issues or submit pull requests.
Built with Rust