FAISS

Facebook AI Similarity Search—high-performance local vector search.

Quick Start

skill-seekers scrape --format faiss --config configs/react.json

Setup

pip install faiss-cpu  # or faiss-gpu

Python Example

import faiss
import numpy as np
import json

# Load data
with open("output/react-faiss.json") as f:
    data = json.load(f)

# Build index
dimension = 1536  # OpenAI embedding size
index = faiss.IndexFlatL2(dimension)

# Add vectors
vectors = np.array([d['vector'] for d in data]).astype('float32')
index.add(vectors)

# Search
query = np.random.random((1, dimension)).astype('float32')
distances, indices = index.search(query, k=5)

print(f"Nearest neighbors: {indices[0]}")

Features

  • ✅ Local & fast - No network latency
  • ✅ GPU support - CUDA acceleration
  • ✅ Multiple indices - IVF, HNSW, etc.
  • ✅ Billions of vectors - Scalable architecture