Building Sovereign In-Memory Vector Search & Neural Reranking in Pure Python: HNSW Graphs, SQ8 Quantization & Hybrid BM25 Fusion

In modern artificial intelligence engineering, Retrieval-Augmented Generation (RAG) and semantic search are standard table stakes. Yet most engineering teams surrender their sovereignty to expensive,