📚 UniMate – AI University Assistant

UniMate is an AI-powered assistant that lets you query your PDF documents in natural language. Upload reports, notes, or resumes; ask a question; and get a clear, well-formatted answer grounded in your files. If the answer is not in your PDFs, UniMate falls back to a high‑quality general answer using Google Gemini.

✨ Features

• 📂 Multi-PDF support across all uploaded documents
• 🔍 Hybrid retrieval (FAISS dense vectors + BM25 lexical)
• 🧠 LLM answers via Google Gemini 1.5‑Flash; text-embedding-004 for embeddings
• 💬 Chat-style Streamlit UI with conversational history
• 📝 Clean formatting with numbered lists, bullets, and inline citations (page numbers)
• 🗂 Topic management with persistent history
• ⚡ Vector index caching for previously uploaded PDFs
• 🔐 Secure .env-based configuration (keys not tracked in git)

🛠️ Tech Stack

• Streamlit – web app framework
• Google Gemini API – LLM + embeddings
• FAISS – dense vector search
• rank-bm25 – lexical retrieval
• PyMuPDF – PDF parsing
• regex – text cleanup and formatting

📦 Project Structure

unimate-ai-assistant/
├─ app.py                   # Streamlit entrypoint
├─ core/                    # Core logic (backend code)
│  ├─ config.py             # Constants, paths, env loader
│  ├─ pdf_utils.py          # PDF parsing + chunking
│  ├─ embeddings.py         # Gemini embeddings
│  ├─ vector_store.py       # FAISS + BM25 hybrid search
│  ├─ llm.py                # Gemini text generation
│  ├─ retrieval.py          # Prompting & context builder
│  ├─ formatting.py         # Answer formatting utilities
│  └─ history.py            # Persistent chat history/topics
├─ backend/                 # Stored data
│  ├─ vector_store/         # Cached FAISS indices
│  └─ history/              # Saved chat threads
├─ data/
│  └─ uploaded_files/       # Uploaded PDFs
├─ requirements.txt         # Python dependencies
└─ .env                     # Your Gemini API key (not tracked in git)

🚀 Getting Started

1) Clone the repository

git clone https://github.com/anshtrivediaiml/UniMate-AI-University-Assistant.git
cd UniMate-AI-University-Assistant

2) Create and activate a virtual environment

python -m venv .venv

On Windows PowerShell:

.venv\Scripts\Activate.ps1

On Linux / macOS:

source .venv/bin/activate

3) Install dependencies

pip install -r requirements.txt

4) Add your API key

Create a .env file in the project root:

GOOGLE_API_KEY=your_gemini_api_key_here

5) Run the app

streamlit run app.py

Then open http://localhost:8501 in your browser.

🔒 Security

• Never commit .env files or API keys.
• The repository’s .gitignore is configured to keep secrets and cache files out of Git.