Full RAG pipeline — semantic chunking, deduplication, vector DB setup. Transform unstructured data into an AI-ready knowledge base.