🧠 GenAI Insurance Document Review Agent

This project is a secure, agentic LLM-based assistant that helps insurance analysts extract clauses, assess policy compliance, and summarize documents using local LLMs. Built for privacy-conscious enterprises like USAA.

---

🚀 Features

• Local LLM support (Mistral 7B, LLaMA3 via Ollama or LM Studio)
• Fine-tuning on insurance-specific clause and QA datasets (with QLoRA)
• Agentic document workflow with LangGraph
• Clause retrieval with Chroma vector DB
• Streamlit front-end for simple document upload and review

---

🧩 Repo Structure

• backend/ – Core logic (parsing, LLM inference, retrieval, agent orchestration)
• fine_tuning/ – Dataset prep and fine-tuning scripts
• frontend/ – Streamlit app UI
• models/ – (Optional) local fine-tuned model folders
• data/ – PDFs and generated Chroma index

---

⚙️ Setup Instructions

# Create environment
python -m venv venv
source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

# Run Streamlit frontend
streamlit run frontend/app.py