@nosql/vector

High-performance vector database package with sub-10ms similarity search for AI/ML workloads.

Features

• Sub-10ms Search - HNSW and IVF indexing for ultra-fast queries
• 1M+ Vectors - Scale to millions of vectors per database
• Multiple Distance Metrics - Cosine, Euclidean, Dot Product, Manhattan, Hamming
• Metadata Filtering - Advanced filtering and hybrid search
• Vector Quantization - 4/8/16-bit quantization for storage efficiency

Installation

```bash npm install @nosql/vector ```

Quick Start

```typescript import { VectorDatabase } from '@nosql/vector';

const vectorDB = new VectorDatabase({ dimensions: 1536, metric: 'cosine' });

await vectorDB.upsert([ { id: 'vec1', vector: [0.1, 0.2, ...], metadata: { text: 'Hello' } } ]);

const results = await vectorDB.search({ vector: queryVector, topK: 10 }); ```

License

MIT