Code Coach
A coding practice platform for a university ACM chapter
coding practice platform · University of Michigan–Dearborn
Code Coach is a practice platform for the University of Michigan–Dearborn ACM chapter. Students prepare for ICPC contests and technical interviews with practice problems, an in-browser editor, code execution, and an AI assistant in one place. It was a three-person senior design project. I designed it, made the logo, and built most of the frontend and backend.
- role
- Design & full-stack development
- with
- University of Michigan–Dearborn
- when
- 2024 – 2025
- stack
- React, AWS Lambda, API Gateway, MongoDB, OpenAI, Judge0
the workspace
Students browse ICPC problems by difficulty, year, and region, or FAANG-style interview questions by category. The problem page puts the statement, an editor, and the assistant side by side. Code runs through Judge0, and the assistant receives the problem and the student's current solution, so its hints are about their code. The platform also has an interview-prep area with a resume builder, discussion forums, and course pages where instructors group problems for a class.
brand identity
I made the Code Coach logo: a mark of two angle brackets around a dot, set beside a lowercase wordmark. The interface uses a minimal palette and works from phones to desktops.
implementation
The frontend is React, deployed to Netlify from GitHub. The backend is serverless: API Gateway routes requests to Lambda functions, and MongoDB Atlas stores users, problems, submissions, and courses through Mongoose. Judge0 runs submitted code in a sandbox and returns its output, runtime, and memory use. The AI help uses OpenAI API via the university's UMGPT service.