AI PDF Chatbot

An AI-powered PDF Chatbot built using FastAPI, Streamlit, FAISS, Sentence Transformers, and Google Gemini AI.

Upload PDF documents and interact with them conversationally using Retrieval-Augmented Generation (RAG), semantic search, and AI-powered contextual responses.

---

Features

• PDF Upload & Processing
• PDF Text Extraction
• Recursive Text Chunking
• Local Embedding Generation
• FAISS Vector Database Integration
• Semantic Search Retrieval
• Gemini-powered Response Generation
• Conversational Chat Interface
• Retrieved Source Chunk Display
• FastAPI Backend
• Streamlit Frontend
• Modular RAG Pipeline Architecture

---

Tech Stack

Backend

• FastAPI
• Python
• Uvicorn

Frontend

• Streamlit

AI / RAG Components

• Google Gemini API
• Sentence Transformers
• FAISS
• LangChain Text Splitters

PDF Processing

• PyMuPDF

---

Project Architecture

PDF Upload
    ↓
PDF Text Extraction
    ↓
Recursive Chunking
    ↓
Embedding Generation
    ↓
FAISS Vector Storage
    ↓
User Query
    ↓
Query Embedding
    ↓
Semantic Retrieval
    ↓
Context Injection
    ↓
Gemini Response Generation

---

Folder Structure

AI-PDF-CHATBOT
│
├── backend
│   ├── app
│   │   ├── core
│   │   │   └── config.py
│   │   │
│   │   ├── routes
│   │   │   ├── upload.py
│   │   │   └── chat.py
│   │   │
│   │   ├── services
│   │   │   ├── chunk_service.py
│   │   │   ├── embedding_service.py
│   │   │   ├── pdf_service.py
│   │   │   ├── rag_service.py
│   │   │   ├── retrieval_service.py
│   │   │   └── vector_store_service.py
│   │   │
│   │   └── main.py
│   │
│   ├── data
│   │   ├── uploads
│   │   └── vector_store
│   │
│   ├── .env
│   └── requirements.txt
│
├── frontend
│   └── streamlit_app.py
│
├── screenshots
├── README.md
└── requirements.txt

---

Installation

1. Clone Repository

git clone https://github.com/decoded15/ai-pdf-chatbot.git
cd ai-pdf-chatbot

---

2. Create Virtual Environment

python -m venv venv

Activate Environment

Windows

venv\Scripts\activate

Mac/Linux

source venv/bin/activate

---

3. Install Dependencies

Backend

cd backend

pip install -r requirements.txt

Frontend

cd frontend

pip install streamlit requests

---

4. Add Gemini API Key

Create a .env file inside backend/

GOOGLE_API_KEY=your_api_key_here

---

Running The Project

Start FastAPI Backend

cd backend

uvicorn app.main:app --reload

Backend runs on:

http://127.0.0.1:8000

---

Start Streamlit Frontend

cd frontend

streamlit run streamlit_app.py

Frontend runs on:

http://localhost:8501

---

RAG Workflow

PDF Upload
      ↓
Text Extraction
      ↓
Chunk Generation
      ↓
Embedding Creation
      ↓
FAISS Vector Indexing
      ↓
User Question
      ↓
Query Embedding
      ↓
Semantic Retrieval
      ↓
Context Injection
      ↓
Gemini AI Response

---

Screenshots

Home

---

PDF Upload & Processing

---

Chat Interface

---

Key AI Engineering Concepts Learned

• Retrieval-Augmented Generation (RAG)
• Semantic Search
• Vector Databases
• Embeddings
• Recursive Chunking
• Prompt Engineering
• Context Injection
• Similarity Search
• AI Orchestration Pipelines
• FastAPI Backend Architecture
• Streamlit Frontend Integration
• Local Embedding Models
• Semantic Retrieval Systems

---

Future Improvements

• Multi-PDF Support
• Conversational Memory
• Citation-based Responses
• Hybrid Search
• Reranking
• OCR Support
• Authentication System
• Cloud Vector Database Integration
• Streaming Responses
• Docker Deployment

---

Author

Built by Dibyansh (decoded15)