Move-Out Agent
Built at Meta Global AI Developer Hackathon · Oct 26, 2026 · Remote

Move-Out Agent helps international students sell their belongings before moving out. A student uploads photos of their room and enters a move-out date. The agent is designed to use a Muse model through the Meta Model API to identify sellable items and return structured data: name, category, condition, a listing description, an ideal price, and a quick-sale price. It asks one short follow-up question only when the answer changes the price (for example, how old a mattress is). It then builds a price-drop schedule from the move-out date, so the plan for three weeks left is different from the plan for two days left, and exports ready-to-post listings. Honest status: the web app, the price-drop logic (calculated in code), spend limits, and an evaluation page are built and deployed. The Edge Function that calls the Muse model is written and tested against a stand-in service. I could not connect the live model: my hackathon credits never arrived and my account was flagged for regional eligibility. The demo therefore uses sample data, and the app labels it clearly. Feicai, my existing student marketplace, is a separate project. Everything in this repository was built during the hackathon. Built with: Meta Model API (Muse), Supabase Edge Functions, Vercel, GitHub.