Use this when showing the project to a recruiter, project lead, professor, or Handshake AI reviewer.
This is EdgeCaseForge AI. It helps design coding benchmark tasks for evaluating AI coding models. Instead of only writing solutions, the app focuses on the harder part: creating problems, hidden tests, edge cases, and model-failure analysis.
- Open the Streamlit app.
- Pick "Room Collision Validator" because it connects to game development.
- Show the "Why this can make AI models fail" tab.
- Explain that many models check only whether the player center is inside a wall, but the correct solution needs circle-rectangle overlap.
- Open the "Test Runner" tab.
- Paste a solution and run sample tests.
I built this project to show that I can think like an AI evaluator, not just a programmer. The app designs coding problems with hidden edge cases, explains where AI models may fail, and validates sample solutions.