🧠 AI Tutor (DEVELOPMENTAL soon to be CortexAI)

An open-source AI-powered tutor built to support student learning through adaptive quizzes, coding help, and ethical guidance.

⚠️ Note: This is an active developmental version of the AI Teaching Assistant.
The system is not yet production-ready and may undergo significant changes. The current version is in the process of being consolidated into jupyterquest.
Contributions, feedback, and testing are welcome!

CortexAI+ is designed to be embedded in data science courses as a virtual TA. It uses large language models (LLMs) to answer coding questions, generate personalized quizzes, and flag ethically sensitive responses—all while helping instructors keep tabs on student engagement and integrity.

---

✨ Features

• 📚 Adaptive Quiz Engine – Dynamically generates questions based on student performance
• 🤖 AI-Powered Tutor – Answers student questions about Python, data science, and statistics
• 🧠 Ethics-Aware AI – Detects ethically sensitive prompts and applies nudges or restrictions
• 📊 Quiz Tracking – Stores quiz scores per student for personalized recommendations
• 🔔 Professor Alerts – Optionally sends Slack alerts when ethical triggers are detected
• 🔍 Similarity Detection – Flags assignment uploads for plagiarism using vector search
• 🧩 RAG-Enhanced Help – Retrieval-Augmented Generation from course materials

---

🚀 Quick Start

1. Clone the repo

git clone https://github.com/your-username/ai-tutor.git
cd ai-tutor

2. Install dependencies

Use venv, conda, or pipenv, then:

pip install -r requirements.txt

3. Set your environment variables

Create a .env file:

OPENAI_API_KEY=your-openai-key
QDRANT_API_KEY=your-qdrant-key
SLACK_WEBHOOK_URL=optional-slack-url

4. Run the app

streamlit run app.py

📁 Project Structure

ai-teaching-assistant/
│
├── .streamlit/                  # Streamlit config (e.g., secrets.toml)
├── assignments/                 # Uploaded assignments (you might want to .gitignore this)
├── chroma_db/                   # Vector DB (likely should be ignored or mounted)
├── course_materials/            # Source documents for retrieval
│
├── src/                         # Source code
│   ├── app.py                   # Entry point (rename best version to app.py)
│   ├── ingest.py                # Document ingestion logic
│   ├── add_documents.py         # (could be merged with ingest.py if similar)
│   ├── utils/                   # Helper modules (if needed)
│   ├── ethics/                  # Ethics-related logging & logic
│   │   └── ethics_log.txt       # (move this or convert to CSV/JSON/log format)
│   └── logs/
│       └── chat_log.txt
│
├── data/                        # Any embedded or uploaded data (to be ignored in .gitignore)
│   └── embedded_sources.csv
│
├── docker/                      # Docker-related files
│   ├── Dockerfile               # Default Dockerfile
│   ├── Dockerfile.prod
│   ├── Dockerfile-no-ollama
│   ├── docker-compose.yml
│   └── docker-compose.prod.yaml
│
├── .gitignore
├── .dockerignore
├── .env                         # .gitignored
├── README.md
├── requirements.txt
└── LICENSE                      # MIT License

---

🛠 Tech Stack

• Python
• Streamlit – For the user interface
• OpenAI / Ollama – LLM backend (supports both OpenAI and local models)
• Qdrant – Vector database for document retrieval & similarity detection
• Slack Webhooks – For instructor notifications (optional)
• GitHub Actions – For autograding and daily imports (planned)

---

📈 Planned Enhancements

• 🔁 Integration with GitHub Classroom and LMS systems
• 🧪 Automatic grading of quizzes and coding assignments
• 🧭 Custom course personas for contextualized tutoring
• 🗂 Dashboard for instructors to monitor student progress

---

🧑‍🏫 Built By

Greg T. Chism, Ph.D.
Assistant Professor of Practice
Coordinator of Data Science Student Engagement
University of Arizona College of Information Science
Website | GitHub

---

📄 License

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