📘 DocuMind AI

Your Intelligent Document Assistant - Query PDFs with the power of AI

---

✨ Overview

DocuMind AI transforms how you interact with PDF documents. Instead of manually scanning through pages of content, simply ask questions in natural language and receive instant, context-aware responses powered by advanced AI models running locally on your machine.

By combining Retrieval-Augmented Generation (RAG) technology with a sleek, modern interface, DocuMind provides an intuitive way to extract insights from complex documents.

🚀 Features

• 📁 Seamless PDF Processing - Upload any PDF document and have it instantly analyzed
• 🔍 Smart Document Chunking - Advanced techniques to break documents into meaningful segments
• 🧠 Semantic Search - Find relevant information based on meaning, not just keywords
• 💬 Natural Language Queries - Ask questions in plain English about your documents
• 🎨 Beautiful Dark Mode UI - Eye-friendly interface for extended reading sessions
• 🏎️ Local Model Execution - All processing happens on your machine for privacy and speed

📋 Prerequisites

• Python 3.7+
• Ollama for running LLMs locally

🛠️ Installation

1. Clone & Navigate

git clone <repository-url>
cd documind

2. Set Up Environment

python -m venv venv

Activation:

• Windows: venv\Scripts\activate
• macOS/Linux: source venv/bin/activate

3. Install Dependencies

pip install -r requirements.txt

4. Configure Models

Make sure Ollama is running with:

ollama pull deepseek-r1:1.5b
ollama pull qwen2.5:1.5b

5. Prepare Storage

mkdir -p document_store/pdfs

🚀 Usage

1. Launch Application

streamlit run rag_deep.py

2. Access Interface

Open your browser and navigate to: http://localhost:8501

3. Workflow

1. Upload - Drag and drop your PDF document
2. Wait - The system will process and analyze the content
3. Query - Ask questions about the document in natural language
4. Review - Get AI-powered responses based on document context

⚙️ How It Works

Stage	Process	Technology
1️⃣ Document Ingestion	PDF uploaded and text extracted	PDFPlumber
2️⃣ Text Processing	Document split into semantic chunks	LangChain Splitters
3️⃣ Vectorization	Text chunks converted to embeddings	DeepSeek Embeddings
4️⃣ Knowledge Storage	Embeddings stored for quick retrieval	In-Memory Vector DB
5️⃣ Query Processing	User question mapped to relevant chunks	Semantic Search
6️⃣ Response Generation	AI generates answer based on context	Qwen2.5 LLM

🧰 Tech Stack

• Frontend: Streamlit
• AI Framework: LangChain
• Model Hosting: Ollama
• PDF Processing: PDFPlumber
• Language Model: Qwen2.5 (1.5B parameters)
• Embeddings: DeepSeek-r1 (1.5B parameters)

📈 Future Improvements

• Multi-document support
• Chat history persistence
• Additional file format support
• Custom model configuration
• Citation and source tracking

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

---

Built with ❤️ for knowledge enthusiasts everywhere