MLBridge

MLBridge is a decentralized marketplace that connects machine learning compute providers with task    requesters on the Base network. The platform enables anyone to submit ML computation tasks and have 

them processed by a distributed network of staked providers, with results verified through          

multi-party on-chain consensus.

Providers stake MLB tokens to offer compute resources and earn rewards for completing tasks.        

Verifiers independently validate results, requiring a 66% consensus threshold before results are

accepted. This ensures accuracy without relying on any single trusted party. Dishonest participants

face slashing penalties, aligning economic incentives with honest behavior.

Key features:

  – Task lifecycle management: create, fund, assign, compute, verify, and complete ML tasks entirely

  on-chain

  – Token-based staking: providers stake a minimum of 1,000 MLB and verifiers stake 100 MLB to

  participate

  – Reward distribution: 70% to providers, 25% to verifiers, 5% protocol fee

  – Reputation system: on-chain scoring tracks participant reliability across completed tasks

  – IPFS integration: task data and results stored off-chain with hashes anchored on-chain

  – Upgradeable architecture: all 9 core contracts deployed behind transparent proxies for safe       

  upgrades

Built with Next.js, wagmi, and RainbowKit on the frontend, with Solidity smart contracts using      OpenZeppelin upgradeable proxy patterns deployed on Base.