esteininger/vector-search

Covers sparse and dense vector extraction techniques, similarity metrics (cosine distance, KNN), and transformer-based embeddings for converting text and multimodal content into vector representations. Includes architectural patterns for production deployments with model versioning, feedback loops, and comparisons across engines like MongoDB Atlas and Pinecone. Addresses domain-specific applications including semantic similarity, question-answering, personalization, and cross-modal file search.

270 stars. No commits in the last 6 months.