ramanareddy-ai/legal-AI-case-management

Legal AI Case Management System

Overview

This project implements an AI-powered legal assistant that:

• Ingests structured medical/legal PDFs
• Uses a Retrieval-Augmented Generation (RAG) pipeline with legal-specific chunking
• Embeds document chunks using OpenAIEmbeddings and indexes with ChromaDB
• Queries those chunks based on natural-language questions
• Generates demand letters using Jinja2 templates
• Serves the solution via an MCP server built with FastAPI

Prerequisites

• Python 3.10+
• Docker & Docker Compose (optional but supported)
• OpenAI API Key (for embeddings + LLM)

Setup

1. Start services (if using Docker):

   docker-compose up -d
   

2. Install Python dependencies:

   pip install -r requirements.txt
   

3. Ingest PDFs and build vector store:

   python app/rag_pipeline.py
   

Running

• Launch the MCP API server:

  uvicorn app.main:app --reload
  

• Generate a demand letter:

  python generate_letter.py
  

  Output saved to generated_letter.txt.

RAG Workflow

1. Chunk PDFs using PyMuPDF
2. Embed text using OpenAIEmbeddings (LangChain)
3. Store vectors in ChromaDB
4. Query using natural language
5. Retrieve top-k chunks
6. Use Jinja2 to fill demand letter template with contextual info

---

Dependencies

• PyMuPDF – Parse PDF files
• openai, langchain – Embedding & LLM pipeline
• chromadb – Vector storage
• jinja2 – Demand letter templating
• fastapi, uvicorn – MCP REST API server

---

Repo Structure

├── app/
│   ├── main.py
│   ├── rag_pipeline.py
│   ├── db.py
│   ├── models.py
├── templates/
│   └── demand_letter.jinja2
├── generate_letter.py
├── requirements.txt
├── README.md

---

✅ Deliverables

• ✅ End-to-end pipeline (PDF → demand letter)
• ✅ RAG-based retrieval with chunking & embeddings
• ✅ MCP server via FastAPI
• ✅ Sample demand letter generated
• ✅ This documentation with architecture & instructions

---

For any questions, feel free to reach out at: ramanadata568@gmail.com