SmartPDF Reader

SmartPDF Reader is an AI-powered chatbot that enables you to chat with your PDF documents. It uses a Retrieval-Augmented Generation (RAG) architecture powered by Google Gemini for intelligent, contextual answers based on your uploaded PDFs — resumes, reports, or academic papers.

Motivation

Reading lengthy PDFs manually is slow and inefficient — especially when you're looking for specific information. SmartPDF Reader transforms PDFs into interactive knowledge bases using a chat interface powered by modern language models and RAG architecture.

🤖 What is RAG?

Retrieval-Augmented Generation (RAG) combines:

• Retrieval: Pulling relevant chunks from your PDFs using semantic search via FAISS and Google embeddings.
• Generation: Feeding that retrieved context into a language model (Google Gemini) to generate an accurate, grounded response.

This enables high-quality, contextual answers that are anchored in your documents.

⚙️ How It Works (RAG Architecture)

flowchart 
    Upload[📄 User uploads PDFs] --> Extract[🔍 Text extracted using PyPDF2]
    Extract --> Split[ Text split into chunks (LangChain Splitter)]
    Split --> Embed[🔡 Chunks embedded using Gemini Embeddings]
    Embed --> Store[ Stored in FAISS Vector DB]
    Question[❓ User asks a question] --> Search[🔎 FAISS similarity search]
    Search --> Retrieve[📚 Retrieve top relevant chunks]
    Retrieve --> Generate[ Gemini LLM processes context + question]
    Generate --> Answer[ Final answer returned to Streamlit UI]

📁 Project Structure

SmartPDF-Reader/
│
├── assets/                  # Screenshots, images
├── faiss_index/             # Vector store folder (generated)
├── .gitignore
├── app.py                  # Main Streamlit UI with RAG logic
├── langchain_rag_chatbot   # Alternate file for RAG logic
├── rag_pipeline            # Extra reference scripts
├── rag_pipelines
├── README.md               # You're reading it!
└── requirements.txt        # All Python dependencies

🖥️ Technologies Used

Component	Tool/Library
Frontend	Streamlit
PDF Parsing	PyPDF2
Text Splitting	LangChain
Embedding	Google Generative AI Embeddings
Vector Store	FAISS
Language Model	Gemini (gemini-1.5-flash)
RAG Backbone	LangChain RAG Chain (load_qa_chain)
UI Styling	Streamlit + Custom HTML/CSS

🚀 Installation & Setup

Prerequisites

• Python 3.8 to 3.11
• A Google Generative AI key

🔧 Setup

git clone https://github.com/yourusername/SmartPDF-Reader.git
cd SmartPDF-Reader

python -m venv venv
source venv/bin/activate  # or venv\Scripts\activate on Windows

pip install -r requirements.txt

Create a .env file in the root folder:

GEMINI_API_KEY=your_google_key_here

▶️ Run the App

streamlit run src/streamlitMain.py

📷 Screenshots

🔥 Features

• Chat with multiple PDFs
• Backed by RAG architecture (retrieval + generation)
• Powered by Google Gemini embeddings + FAISS
• Download chat as CSV
• Beautiful dark-themed UI

🛠️ Future Improvements

• Add local/offline LLM support via Ollama
• Summarization feature
• Multi-language support
• Chat export as PDF

👩‍💻 Developed By

Shafia Saba

• 🌐 GitHub
• 🌐 Gitlab
• 📧 Email: shafiasaba2004@email.com