The Problem
Virginia Tech is currently in a housing crisis where students can't get on-campus housing, so the administration is pushing everyone to lease off campus. The whole process is full of questions and loopholes that leasing companies exploit, leaving students with no good options.
That's where Bed-a-thon comes in.
The Solution
We built a platform that makes finding compatible roommates and securing housing as simple as possible. Students answer questions about their lifestyle, budget, location preferences, and amenities. Our matching algorithm converts those answers into preference vectors and uses Euclidean distance to score compatibility between potential roommates. Light weights distinguish "nice-to-have" from "must-have" criteria, so matches actually make sense.
Once a group forms, the system recommends apartments that fit the group's combined needs. But here's the thing: we didn't stop at recommendations. Our AI agent actually calls the leasing offices for you. It schedules tours, asks the common questions (pricing, availability, utilities, lease terms), and when appropriate, negotiates for specials. We found that most of these leasing companies are not transparent about their policies and are not willing to negotiate unless students call or visit in person.
We're really proud of how well the AI agent performs at conversing with leasing offices. It gets the information students need—tour times, price breakdowns, lease details—without manual phone calls.
Tech Stack
We built the frontend with Next.js, React, TypeScript, and Tailwind for a clean, responsive interface. The backend runs on Python FastAPI with SQLModel ORM for type-safe APIs. We use Supabase for PostgreSQL database with real-time sync, so matches update instantly, plus Supabase Auth with Google OAuth for seamless sign-in.
Thoughts
The biggest challenge? None of us had used Supabase before, so we had to learn Auth, Realtime, and database features on the fly. We also hit some frontend-backend connection issues mid-hack, but fixed them by simplifying API responses and adding stricter validation.
Right now we're working on expanding to other campuses and adding clearer "why you matched" explanations, calendar integrations for auto-booking tours, and basic safety features like verified profiles and reporting.