The problem
Dorm life runs on small favors: borrowing a charger, an iron, a pot, a ride. On a campus of a thousand-plus dormers those favors happen in group chats where a request scrolls out of sight in minutes, and where you have no idea whether the person answering is reliable.
The constraint
The interesting problem in peer-to-peer sharing is not inventory. It is trust and follow-through between people who are strangers but also neighbors, and who will see each other again. A listings page does not solve that. What breaks a sharing network is a request nobody answers, or a handoff that goes wrong with no record of it.
The decision
I directed a three-person team and owned the backend, which I built as three connected systems rather than one feature list. A requests-and-offers tracker with live notifications, so a request has a state instead of scrolling away. A chat system specialized for the handoff itself, rather than a general messenger bolted on. And a user review framework, so follow-through accumulates into something a stranger can read before agreeing to meet.
The outcome
MakeAbot serves over a thousand dormers. The part I would defend in a review is the decision to treat reviews and notification state as core backend concerns from the start, rather than as engagement features to add later. In a network of neighbors, reputation is the product.
Screenshot
