This repository implements 3 different types of search systems: traditional full-text search, vector search and hybrid search.

Keyword search over public‑domain books using Whoosh.

• Data download + normalization pipeline
• BM25F scoring with field boosts
• CLI + Streamlit UI

See: book-search/README.md

Image and text search over a small image set.

• OpenCLIP embeddings stored in DuckDB
• Query by text or image
• CLI + Streamlit UI

See: image-search/README.md

Hybrid search for support tickets and KB docs.

• BM25 (Whoosh) + vector search (DuckDB vss)
• Optional cross‑encoder reranking
• CLI + Streamlit UI

See: support-ticket-search/README.md

Each project is standalone. Create a virtualenv and follow its README:

• book-search/README.md
• image-search/README.md
• support-ticket-search/README.md

Run these three commands (one per terminal):

cd support-ticket-search && streamlit run app.py --server.port 8501
cd book-search && streamlit run app.py --server.port 8502
cd image-search && streamlit run app.py --server.port 8503