Fullstack Gen AI PDF Context-Aware Chatbot

Overview

This is a full-stack Gen AI application that functions as a PDF context-aware chatbot. Users can upload PDF documents and ask questions related to their content. The application leverages a Large Language Model (LLM) to provide relevant answers based on the PDF context.

Setup

For Backend:

1. Navigate to the backend directory:

   cd backend

2. Create a virtual environment:

   python -m venv venv

3. Activate the virtual environment:
   • Windows:

     venv\Scripts\activate

   • Linux:

     source venv/bin/activate

4. Install the required dependencies:

   pip install -r requirements.txt

5. Run the FastAPI application:

   uvicorn main:app --reload --host 0.0.0.0 --port 8000

For Frontend:

1. Navigate to the frontend directory:

   cd frontend

2. Install the npm dependencies:

   npm install

3. Start the development server:

   npm run dev

API Endpoints

/upload (POST)

• Receives files via a POST request.
• Extracts text content from the uploaded PDF using pymupdf.
• Divides the extracted text into smaller chunks.
• Generates vector embeddings for each chunk using Langchain.
• The text extraction and embedding processes are executed in separate threads to handle large files and CPU-intensive tasks efficiently.
• Stores file metadata (e.g., filename, upload timestamp) in a SQLite database.
• Saves the generated vector embeddings in the Fasis vector database.

/ask (POST)

• Retrieves relevant vector embeddings from the Fasis index based on the user's question.
• Receives the user's question via the POST request body.
• Combines the retrieved context (from the PDF) and the user's question.
• Sends this combined information to the LLM.
• Returns the LLM's response to the user.

Tools and Frameworks

• Frontend:
  • Vite: A fast build tool and development server.
  • Tailwind CSS: A utility-first CSS framework.
  • React: A JavaScript library for building user interfaces.
• Backend:
  • FastAPI: A modern, high-performance web framework for building APIs with Python.
  • PyMuPdf: A Python library for working with PDF and XPS documents.
  • Langchain: A framework for developing applications powered by language models.
• Database:
  • Fasis: A vector database for efficient similarity search.
  • SQLite: A lightweight, disk-based database for storing metadata.

Demo Video

AppDemo.mp4