Things I've built, mostly vibecoded. Each one starts with a real person's situation and ends with something they can use.
AI group-travel matching with group-size scoring. I found that trust is what stops groups from booking, shipped three changes after prototype testing, and scoped a phased rollout to cut skipped trips from 74% to under 50%.
Won 1st of 44 teams, 250+ participantsUSC Marshall MSBA AI Case Competition.
Open TripMatch ↗ View on GitHub ↗
A bid planner for MSBA electives. It lets you filter sections by day, time and open seats, tracks your units against degree requirements, flags time conflicts and helps you split your bid points across classes.
Open the planner ↗ View on GitHub ↗
I work in a team of five to seven students with a staff sponsor who owns a real process. Over the cycle, we:

A privacy-first digital jigsaw site. Photos become puzzles with no server storage, and progress saves from one puzzle to the next. I reworked piece sorting and drag-and-drop from playtester feedback, then added sorting trays, difficulty guides and progress tracking.

An AI quiet-spot finder for overwhelmed students between classes. I defined the target user and core problem, prioritized the MVP filters, curated about 40 spots and built the monetization model.