MedIntel is a production-grade AI assistant that answers biomedical questions using a fine-tuned SciBERT model and Retrieval-Augmented Generation (RAG) pipeline over PubMed literature.
- Fine-tuned SciBERT for biomedical question classification (
yes/no/maybe) - Retrieval-Augmented Generation (RAG) using FAISS + MiniLM + PubMed
- Streamlit frontend for real-time interactive QA
- Dockerized and production-ready
- Clean architecture and modular code
pip install -r requirements.txt
2. Run the App (Local)
streamlit run streamlit_app/app.py
3. Run with Docker
docker build -t medintel-app .
docker run -p 8501:8501 medintel-app
python train_classifier.py
python rag/index_pubmed.py
Author
Chetan Nani
Built with 💡 to showcase end-to-end AI product engineering in healthcare.
---
Once this is saved and committed, reply `done`, and I’ll be moving to **cloud deployment (Render or Hugging Face)** so you can showcase it instantly with a public link.