Ebby
OIT Track winner and MLH Best Use of PropelAuth, HackSMU VI
A student well-being app: an AI health assistant, anonymous peer support groups, and webcam exercise tracking that counts your reps.
- Role
- Team, HackSMU VI (2024). My part: the Next.js app, PropelAuth login, the design, the support-groups flow and questionnaire, and the exercise pages.
- Stack

Problem
Students juggle mental and physical health on their own, and campus resources are scattered across pages nobody reads. We wanted one place to ask questions, find people going through the same thing, and keep moving.
What I built
Ebby, a chat assistant for wellness questions. Peer support groups: a short questionnaire recommends groups, and members chat anonymously through TalkJS. Exercise tracking: the webcam stream goes over a WebSocket to a Flask server that uses MediaPipe pose landmarks and joint angles to count squats and bends.
My part (top committer): I set up the Next.js app and its design system, added PropelAuth login and route protection, built the questionnaire and the support-groups pages (group recommendations, joining, and the anonymous TalkJS group chat), and reworked the exercise pages and rep-counter UI. Caleb Bae built the pose-detection server and sachi-jh the chat assistant.
Results
- Won the OIT Track and MLH’s Best Use of PropelAuth at HackSMU VI.
- We also scraped university resources into a retrieval index for the assistant, but didn’t get it wired into the frontend before judging.
