GitHub Repository Search & Data Pipeline

Personal repository-discovery project combining a collection/enrichment pipeline with multilingual local search. The pipeline deduplicates repository metadata and README content with resumable collection. Search combines SQLite FTS5/BM25 keyword retrieval and Qdrant embedding search using reciprocal rank fusion, with optional cross-encoder reranking, filters, CLI, and Streamlit interface. Scope: Grouped data/search components, not separate product counts. Source inspection establishes implemented paths, not independently measured retrieval quality or latency.