This project demonstrates how to build a Retrieval-Augmented Generation (RAG) pipeline using clinical documents from the MIMIC-IV-Extended dataset. It leverages domain-specific embeddings and the Ollama inference server (e.g., Mistral) to answer clinical queries.