Padly is a roommate-first housing platform for students, interns, and early-career professionals.
Most rental apps start with listings. Padly starts with people.
Finding housing with roommates is a two-sided problem: you need the right people and the right place. Traditional platforms only solve one side, leaving users to figure out the other on their own.
New users go through onboarding to define their hard requirements (budget, location, move-in date) and soft preferences (lifestyle, cleanliness, schedule). These drive every recommendation that follows.
Padly suggests roommates based on preference compatibility, lifestyle alignment, and behavioral signals. Users can express interest in each other through an intro system with mutual opt-in.
Compatible users form groups. Each group has shared preferences automatically aggregated from its members. As members join or leave, the group profile updates.
The Discover page presents listings ranked for the group using a blend of:
- Rule-based filtering — hard constraints like budget, location, and bedroom count
- Behavioral signals — learned from swipe interactions (likes, passes, saves)
- Neural ranking — a two-tower model that embeds user preferences and listing features into the same vector space to score affinity
Users swipe through listings and can save favorites to their group.
Padly runs a stable matching algorithm that pairs groups with listings, producing ranked matches where both sides are considered. Groups see their matches with explainable scores.
Every interaction feeds back into the system. Swipe behavior refines recommendations over time, so the more a group uses Padly, the better the suggestions get.
- Guided onboarding with a walkthrough tour for new users
- Roommate suggestions with compatibility scoring and intro requests
- Group management with invitations, join requests, and shared preferences
- Swipe-based discover for browsing recommended listings
- Group saves to bookmark listings for the whole group
- Stable matching between groups and listings
- Metro-aware location matching across cities and regions
- Row-level security on all database tables
- Frontend: Next.js 15, React 19, Mantine UI, TanStack Query
- Backend: FastAPI, Supabase (Postgres + Auth), Pydantic
- AI/ML: TensorFlow two-tower model, behavioral fingerprinting, embedding-based similarity
- Infrastructure: Supabase migrations, Git LFS for model artifacts
- Students moving for school terms
- Interns relocating temporarily
- New grads and early-career professionals moving to new cities
- Anyone who needs both a home and compatible roommates