PRODUCTION AI INFRASTRUCTURE · SECURITY
CS 220 AI Learning Infrastructure
I led a 12-TA team and built a full-stack AI coding tutor for 900+ students: a TypeScript JupyterLab extension, Azure OpenAI, FastAPI/MySQL and Redis services on an Azure Linux VM, plus a React dashboard and test-gated releases.Built an AI tutor for 900+ students, saved $2,900 in inference costs, and led a 12-TA team.
AI coding tutor used by 900+ students
Lower costs for the course
Faster support for student issues
01PROJECT THREAD
A coding tutor for the whole course
I built and launched a full-stack AI coding tutor for more than 900 students, connecting a TypeScript JupyterLab extension to Azure OpenAI through FastAPI, MySQL and Redis. Docker Compose ran the services on an Azure Linux VM.
02PROJECT THREAD
Lower inference costs, useful usage data
Caching and per-IP rate limiting cut inference spend from $3,500 in fall to $600 in spring. I also built a React/TypeScript/Node.js dashboard so instructors could understand how students used the tutor on each assignment.
03PROJECT THREAD
Turning issue reports into tested fixes
AI-agent automation triaged GitLab issues, reviewed code and opened test-gated merge requests, cutting student-reported issue resolution from days to minutes.
04PROJECT THREAD
Leading the engineering team
I led a 12-TA team, coordinated parallel work on programming projects and course infrastructure, reviewed merge requests and set up developer tooling with Pixi.