A Pythonic vector database designed for simplicity and scalability, offering core CRUD operations and flexible deployment options from local to cloud environments. It targets developers needing a lean yet powerful solution for managing and searching vector embeddings, leveraging DocArray for search logic and Jina for scalable index serving.

How It Works

vectordb utilizes DocArray as its core engine for vector search logic, enabling efficient Approximate Nearest Neighbor (ANN) and Exact Nearest Neighbor (ENN) searches. Jina provides the underlying infrastructure for scalable index serving, supporting sharding and replication for high availability and throughput. This architecture allows vectordb to function as a standalone library or be served as a scalable service via gRPC, HTTP, or WebSockets.

Quick Start & Requirements

• Install: pip install vectordb
• Prerequisites: Python 3.x, NumPy. HNSWVectorDB requires HNSWLib.
• Setup: Local setup involves defining a BaseDoc schema with DocArray and initializing a database class (e.g., InMemoryExactNNVectorDB, HNSWVectorDB).
• Docs: https://docs.jina.ai/concepts/vectordb/

Highlighted Details

• Offers both Exact NN (InMemoryExactNNVectorDB) and Approximate NN (HNSWVectorDB) search capabilities.
• Supports serving as a service via gRPC, HTTP, and WebSocket protocols.
• Provides sharding and replication for scalability and availability.
• Integrates with Jina AI Cloud for seamless cloud deployment.

Maintenance & Community

• Backed by Jina AI.
• Community support via Discord: https://discord.com/invite/kN73sdX
• Contributions are welcome.

Licensing & Compatibility

• License: Apache-2.0.
• Compatibility: Permissive license suitable for commercial use and integration with closed-source applications.

Limitations & Caveats

Currently, Jina AI Cloud deployments are limited to 1 replica; support for N replicas in the cloud is under development. The roadmap indicates plans for more ANN algorithms and enhanced filtering capabilities.