Two-Stage Retrieval Pipeline

Implement a two-step RAG pipeline by first running an embeddings-based similarity search to get a pointer, then executing a SQL or graph query to fetch the full detailed dataset.