AcadRAG – Academic Retrieval-Augmented Generation Assistant

AcadRAG is a full-stack, AI-powered academic assistant that converts study PDFs into a persistent, interactive knowledge base. It allows users to upload, manage, and query academic documents using natural language, delivering accurate, context-aware answers through a Retrieval-Augmented Generation (RAG) pipeline — all running locally and privacy-first.

Features

• User Authentication: Secure login and registration system with user-isolated data.
• Document Management (CRUD): Upload, view, and delete academic PDFs per user.
• Multi-PDF Support: Processes multiple academic documents together for richer context.
• Text Extraction & Cleaning: Uses 'pdfplumber' to extract clean, structured text from PDFs.
• Chunk-Based Indexing: Splits large documents into smaller chunks for efficient retrieval.
• FAISS Vector Search: Enables fast semantic search over indexed document embeddings.
• Local AI Model Integration: Works with Ollama models (e.g., gemma:2b-instruct) for question answering.
• Context-Aware Q&A: Combines retrieved document chunks with LLMs to generate detailed academic answers.
• Offline & Privacy-Friendly: Runs fully offline using Ollama; no data leaves the system.
• SaaS-Ready Architecture: Modular Flask-based design suitable for future cloud deployment.

Project Structure

ACADRAG
│
├── app.py                  # Flask application entry point
├── acad_rag.py              # Core RAG pipeline (ingest + query)
│
├── routes/
│   └── documents.py         # Document routes (upload, list, delete)
│
├── services/
│   └── document_service.py  # Document storage & indexing logic
│
├── templates/
│   ├── dashboard.html       # Main Q&A interface
│   ├── documents.html       # Document management UI
│   ├── login.html
│   └── register.html
│
├── static/                  # Static assets (CSS/JS if extended)
├── data/                    # User PDFs & FAISS indexes
│
├── users.db                 # SQLite user database
├── requirements.txt
├── README.md
├── .env.example
└── .gitignore

Getting Started

# Install dependencies
pip install -r requirements.txt

# Install Ollama (https://ollama.ai) and pull a model
ollama pull gemma:2b-instruct
ollama serve

# Run the AcadRAG web application
python app.py

# Open in browser
# http://127.0.0.1:8501

Usage Flow

1. Register / Login via the web interface
2. Upload academic PDFs from the Documents page
3. Ask questions from the Dashboard
4. Receive context-aware answers generated using RAG

Privacy & Offline Support

• All documents remain on the local system
• No cloud dependency required
• Ideal for academic, institutional, and exam-prep use cases
• Production-ready & SaaS-extensible