Semantic Search & Match (Node.js + Python)

Deployed vector search with sentence embeddings and hybrid BM25 for relevance. Built a RAG layer for explainable results, boosted click-through by 28%, and reduced zero-result queries by 62%.