Medical Chatbot — Retrieval-augmented Q&A

A retrieval-augmented medical assistant that surfaces relevant health information through semantic search and context-aware generation. The focus was turning a knowledge base into a usable chat interface, with a containerized deployment path to AWS.

People want quick, understandable medical information, but in a high-trust domain answers need to stay tied to source material.

Semantic retrieval combined with LLM generation, so answers are informed by retrieved domain content.

A Flask app serves the chat, Pinecone stores the vector index, LangChain orchestrates retrieval, and OpenAI generates answers over the retrieved context.

• Built the full workflow from embedding generation to chat interface.
• Used Pinecone retrieval so answers are based on indexed source content.
• Explored containerized deployment and CI/CD-oriented hosting on AWS.

• Balancing usefulness with the sensitivity of the domain.
• Keeping retrieval quality high across varied medical questions.
• Reducing latency without oversimplifying the pipeline.

Python · Flask · LangChain · Pinecone · OpenAI GPT · Docker · AWS EC2