📄 RAG File QA Chatbot

Chat with your PDFs using LangChain + Streamlit + OpenAI

---

📌 Overview

This project is a Retrieval-Augmented Generation (RAG) chatbot built with Streamlit + LangChain + OpenAI.
It allows you to upload PDF files, query them in natural language, and get AI-powered answers with sources.

• 📄 Upload multiple PDFs
• ✂️ Split text into smart chunks
• 🧠 Generate embeddings with OpenAI
• 💾 Store + retrieve chunks using ChromaDB
• 🤖 Ask questions & get contextual answers with sources shown

---

🚀 Recent Improvements & Fixes

We’ve made several significant upgrades and fixes to evolve this chatbot from a simple prototype to a robust, production-ready conversational application.

🧠 1. Added Conversational Memory

• Before: Each query was treated independently. Follow-ups like “why?” made no sense.
• After: Implemented a history-aware RAG chain that reformulates follow-ups into full contextual questions (e.g., “why is the sky blue?”).
  ➤ The chatbot now maintains multi-turn memory, enabling natural and contextually aware conversations.

⚡ 2. Modernized Streaming Engine for Streamlit

• Before: Used outdated callbacks (StreamHandler) that caused NoSessionContext errors.
• After: Replaced with LangChain’s modern .stream() method.
  ➤ This allows real-time response streaming directly in Streamlit without threading issues.

🧰 3. Fixed Code-Level Bugs

• 🔄 Circular Import Fixed: Renamed local langchain.py → app.py to avoid module conflicts.
• 🔕 Disabled ChromaDB telemetry logs for cleaner terminal output.

In summary, your chatbot is now stable, conversational, and fully interactive, ready for production use.

---

✨ Features

• 📂 Multi-PDF Support — Upload and query multiple documents
• 🧩 Chunking & Embedding — Splits content for better context retrieval
• 🔍 RAG Pipeline — Retrieval + Context-aware AI answers
• 🧠 Conversational Memory — Handles follow-up questions seamlessly
• 💬 Real-time Streaming — Smooth token-by-token response in Streamlit
• 📊 Source Transparency — Displays top 3 document sources
• ⚡ Streamlit UI — Simple and interactive interface

---

🛠️ Tech Stack

Layer	Technologies	Purpose
Frontend	Streamlit	Interactive UI
Backend	Python, LangChain	RAG pipeline & orchestration
Vector DB	ChromaDB	Store & retrieve embeddings
Document Loader	PyMuPDF	Parse PDF files
LLM + Embeddings	OpenAI (GPT + embeddings)	Contextual QA

---

⚙️ How It Works (RAG Pipeline)

1. 📥 Upload PDFs — User uploads documents via Streamlit UI
2. ✂️ Text Splitting — Documents are chunked into smaller passages
3. 🔑 Embedding — Each chunk is embedded using OpenAI embeddings
4. 💾 Vector Store — Chunks + embeddings stored in ChromaDB
5. ❓ Query — User asks a question
6. 🔍 Retriever — Relevant chunks are retrieved
7. 🤖 LLM Response — GPT answers using retrieved context
8. 📑 Sources — Top 3 supporting chunks shown

---

🧪 Local Development

🔧 Requirements

• Python 3.9+
• OpenAI API Key

---

🏁 Getting Started

1. Clone & Setup

git clone https://github.com/your-username/rag-file-chatbot.git
cd rag-file-chatbot

2. Install Dependencies

pip install -r requirements.txt

3. Add API Key

Create a .env file (see .env.example):

OPENAI_API_KEY=your_openai_key

---

🚦 Run the App

streamlit run app.py

The app will run locally at 👉 http://localhost:8501

---

📁 Folder Structure

rag-file-chatbot/
├── app.py              # Main Streamlit app
├── requirements.txt    # Python dependencies
├── .env.example        # Example API keys
├── .gitignore
└── README.md

---

🙌 Acknowledgments

• LangChain
• Streamlit
• ChromaDB
• OpenAI
• PyMuPDF

---

Built with ❤️ by Kartik Garg