Dreamlend Finance

The problem

Pooled lending protocols set terms algorithmically from utilization, which is efficient but gives neither side any say in the deal. Peer-to-peer lending restores that control — a lender writes the offer they actually want — at the cost of a much harder user experience problem, because now every loan is a distinct object a user has to understand.

How it was built

Built as a peer-to-peer lending protocol where lenders create explicit loan offers and borrowers take them, rather than both sides interacting with a pool at rates set by a curve.

The design priority was making trustless lending legible. A P2P protocol exposes far more surface to the user than a pooled one — terms, collateral, duration, counterparty state — and the interface has to make an offer comprehensible at a glance or the model does not work regardless of the contracts underneath.

Solidity contracts developed with Foundry, which brings fuzzing and invariant testing to a domain where the properties that matter — a borrower cannot withdraw collateral backing an active loan, a lender cannot be shortchanged on repayment — are exactly the kind that property-based tests catch and unit tests miss.

GraphQL for indexed on-chain state, since listing and filtering open loan offers is a query problem that direct contract reads handle badly.

Why this stack

• Foundry — fuzz and invariant testing for lending logic where edge cases carry real value.
• GraphQL over an indexer — offer discovery and filtering, which contract calls cannot serve efficiently.

Outcome

• Built as a hackathon collaboration; deployed on Polygon.