💊 DrugGPT

A simple Medical RAG (Retrieval-Augmented Generation) chatbot built using LangGraph, LangChain, Pinecone, Groq, FastAPI, and Streamlit.

This project started as a way to learn how modern AI assistants actually work behind the scenes instead of just calling an LLM API and hoping for the best.

The idea is simple:

• Store medical books in a vector database
• Retrieve relevant information based on the user's question
• Feed that information to an LLM
• Generate grounded answers instead of hallucinating

---

Why this project?

Most AI chatbots can answer medical questions, but they often rely only on their pre-trained knowledge.

DrugGPT tries to answer questions using information retrieved from medical books before generating a response.

This project is also a playground for learning:

• RAG Pipelines
• LangGraph
• Agentic AI
• Vector Databases
• LLM Engineering
• Open Source Collaboration

---

Current Features

• PDF document ingestion
• Medical book processing
• Text chunking
• Embedding generation
• Pinecone vector database
• Semantic search
• RAG pipeline
• LangGraph workflow
• FastAPI backend
• Streamlit frontend
• LangSmith tracing
• Conversation memory

---

Tech Stack

AI

• LangChain
• LangGraph
• Groq (Llama 3.1)
• Sentence Transformers

Backend

• FastAPI

Frontend

• Streamlit

Database

• Pinecone Vector Database

Observability

• LangSmith

Document Processing

• PyMuPDF

---

Project Structure

DrugGPT/
│
├── RAG_pipeline/
│   ├── graph.py
│   ├── rag.py
│   ├── retriever.py
│   ├── vector_db.py
│   ├── state.py
│   ├── main.py
│   ├── frontend.py
│   └── requirements.txt
│
├── .env.example
├── README.md
└── .gitignore

---

Setup

Clone the repository:

git clone https://github.com/your-username/DrugGPT.git

cd DrugGPT

Create a virtual environment:

conda create -n ai python=3.11

conda activate ai

Install dependencies:

pip install -r RAG_pipeline/requirements.txt

Create your .env file:

cp .env.example .env

Add your API keys.

---

Run the Backend

cd RAG_pipeline

python -m uvicorn main:app --reload

---

Run the Frontend

streamlit run frontend.py

---

Roadmap

Phase 1

• Basic Medical RAG
• Pinecone Retrieval
• FastAPI Backend
• Streamlit UI

Phase 2

• Streaming Responses
• SQLite/Postgres Memory
• Resume Conversations

Phase 3

• Multi-Agent System
• Drug Search Tool
• Symptom Analysis Tool
• Medical Calculator Tool
• Web Search Integration

Phase 4

• Multi-modal Inputs
• Medical Reports
• Image-based Diagnosis Research

---

Open Source Contributions

This project is student-friendly.

If you are:

• Learning LangGraph
• Learning RAG
• Learning LLM Engineering
• Looking for your first open-source contribution

you are welcome here.

Some beginner-friendly contribution ideas:

• Improve UI
• Better prompts
• Add citations
• Add memory
• Improve retrieval quality
• Add evaluation metrics
• Improve documentation

No contribution is too small.

---

Disclaimer

This project is for educational and research purposes only.

It is NOT a replacement for professional medical advice, diagnosis, or treatment.

Always consult qualified healthcare professionals for medical decisions.

---

Built By Students, For Students

If you're learning AI Engineering, Data Science, MLOps, or Generative AI and want to contribute, feel free to open an issue or submit a pull request.

Let's build cool stuff and learn together 🚀