ALMA PhD Trial Task: Collaborative LLM Prototype

This repository contains the source code for Part 1 of the ALMA PhD trial task. It implements a "Collaborative Partner" application designed to shift Human-LLM interaction from passive consumption to active collaborative learning.

The prototype utilizes the Google Gemini API to demonstrate pedagogical behaviors such as constructive friction, reflective questioning, and role adaptation (Novice vs. Professional).

📂 Project Structure

📄 Part 1:

• elheyba.ipynb: The main Jupyter Notebook containing the application prototype, system architecture, and design rationale.

📄 Part 2: The Reflection Report

• report.pdf: The accompanying theoretical report.
  • Content: Defines the learning scenario (High-Level Software Architecture), analyzes the "Oracle Trap," and provides the pedagogical justification for the design choices (e.g., why the AI refuses to answer "lazy" prompts).

Note: Detailed explanations of the system architecture and the educational design choices are embedded directly within the notebook's Markdown cells.

🛠️ Requirements

• Python: 3.10 or higher
• API Key: A valid Google Gemini API Key (Free Tier supported)

Dependencies

• google-generativeai (Official Google Gemini Python Client)
• ipykernel (If running inside a Jupyter environment)

🚀 Setup & Usage

1. Clone the repository:

2. Install the dependencies:

   pip install google-generativeai

3. Run the Notebook: Open elheyba.ipynb in Jupyter Lab, Jupyter Notebook, or VS Code.

4. Configure Authentication: The first cell of the notebook will prompt you to enter your Google API Key. Follow the instructions inside the notebook to obtain one if you haven't already.

📜 License

Submitted as part of the ALMA Project PhD recruitment process.