ClosedAI
Store and validate AI training data without a centralized storage. Closed to regulators, open to everyone else.
Project Description
We built a platform dedicated to decentralizing AI training data. We wanted to create a decentralized alternative to centralized image collections for AI training like Imagenet. A contributor first sends the image to a validator through peer to peer data transfer. The validator reviews the contribution, and if they approve, the image is stored in a decentralized storage. In the future, we would like to compensate contributors through smart contracts, guaranteeing fair and prompt payment.
How it's Made
We use lighthouse storage/filecoin as the decentralized storage. We also use PeerJS to enable peer to peer data transfer. We use web3 to connect to Metamask wallets. We used NextJS for frontend and backend. We used a stock background video and CSS hacks to create the landing page hero section.
