📄 AI Document Assistant (Context-Aware RAG)

An intelligent, full-stack document analysis tool powered by Retrieval-Augmented Generation (RAG). This project allows users to upload PDF or DOCX files and have a context-aware conversation with them. It uses a Hybrid AI approach (Local Embeddings + Cloud LLM) to ensure speed, stability, and zero cost.

---

---

🚀 Key Features (Updated)

• 🗣️ Conversational Memory: The AI remembers previous context in the chat, allowing for natural follow-up questions (e.g., "Who is the author?" -> "How old is he?").
• 🔍 Source Citations: Every answer comes with expandable source references, showing exactly which file and page the information was pulled from.
• 📚 Smart Parsing: Uses pdfplumber to accurately read complex PDF layouts, tables, and unconventional fonts.
• ⚡ Hybrid AI Engine:
  • Embeddings: Runs locally using HuggingFace (all-MiniLM-L6-v2) — Unlimited & Private.
  • LLM: Powered by Google Gemini (Flash/Pro) — Fast & High Quality.
• 🐳 Dockerized: Fully containerized with Docker and Docker Compose for easy deployment.
• 🧹 Auto-Reset Memory: Automatically clears the vector database when a new file is uploaded to prevent data contamination.
• 📝 Grammar Analysis: A dedicated tool to check and correct grammar in texts.

🛠️ Tech Stack

Component	Technology	Description
Language	Python 3.11	Stable version for AI/ML libraries.
Backend	FastAPI	REST API to handle file uploads & logic.
Frontend	Streamlit	Interactive Chat UI with history state.
Orchestration	LangChain	Framework for RAG & Contextual Rephrasing.
Vector DB	ChromaDB	Local database to store text embeddings.
Parsing	PDFPlumber	Advanced PDF text extraction.
Deployment	Docker	Containerization for reproducible builds.

---

📦 Installation & Setup

You can run this project either Manually (Local Environment) or using Docker.

Option 1: Manual Setup (Local)

1. Clone the Repository

   git clone https://github.com/YOUR_USERNAME/rag-doc-assistant.git
   cd rag-doc-assistant

2. Create Virtual Environment (Python 3.11 Recommended)

   # Windows
   py -3.11 -m venv venv
   .\venv\Scripts\activate

3. Install Dependencies

   pip install -r backend/requirements.txt
   pip install -r frontend/requirements.txt

4. Configure API Key Create a .env file in the root directory:

   GOOGLE_API_KEY=AIzaSyDxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

5. Run the App (Two Terminals Needed)

   • Terminal 1 (Backend): uvicorn backend.app.main:app --reload
   • Terminal 2 (Frontend): streamlit run frontend/app.py

---

Option 2: Docker Setup (Recommended) 🐳

Ensure you have Docker Desktop installed and running.

1. Configure API Key Create a .env file in the root directory with your GOOGLE_API_KEY.

2. Build and Run

   docker-compose up --build

3. Access the App Open your browser at: http://localhost:8501

---

🔮 Project Roadmap & Status

Feature	Status	Description
RAG Core	✅ Completed	PDF Upload + Vector DB + LLM Answer.
Grammar Check	✅ Completed	Separate tab for linguistic analysis.
Conversational Memory	✅ Completed	Backend rephrases queries based on history.
Source Citations	✅ Completed	UI displays page numbers for verification.
Docker Support	✅ Completed	Dockerfile and docker-compose added.
Smart PDF Parsing	✅ Completed	Integrated pdfplumber for better OCR-like reading.
Multi-File Support	🚧 Planned	Allow querying multiple PDFs simultaneously.
Table Extraction	🚧 Planned	Extract data tables to CSV/Excel.
Cloud Deployment	🚧 Planned	Deploy to Streamlit Cloud / Railway / AWS.

---

🤝 Contributing

Contributions are welcome! Please fork the repository and submit a Pull Request.

📄 License

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