FVM Intro: Build an AI Dataset Marketplace
summary
Overview of building an AI dataset marketplace on Filecoin covering concrete pieces: network scale 6.4 exabytes and 2,600 storage providers with 130 million NFTs stored, onboards petabytes/day, verifiable storage via proof-of-replication and proof-of-spacetime, deal flow using CAR files, Boost, Lotus, FM (Filecoin Virtual Machine), and retrieval via Lighthouse or web3.storage. Demo app remains unfinished but proposed flow shows using Lighthouse API to upload, FM/solidity contracts to automate deals, optional encrypted storage plus off-chain compute for query-only access, and three hackathon prizes for implemented projects.
AI-written summary · about
chapters (17)
- 0:00Welcome
- 1:09Web centralization and AI data
- 3:36IPFS reverses centralization
- 5:21Filecoin for decentralized storage
- 7:11Global Filecoin storage network
- 8:59Filecoin’s permissionless design
- 10:01Public AI datasets on Filecoin
- 11:06Proof mechanisms for storage
- 13:14Filecoin for AI and ML
- 14:57Introducing the Filecoin Virtual Machine
- 16:03How files are stored
- 17:51Automating storage deals
- 18:56Developer tools for Filecoin
- 21:07Decentralized databases for AI
- 22:10Building AI dataset marketplace
- 25:06Hackathon prizes
- 27:13Resources for developers
related talks
26mCatching Up with the New Data Economy
28mOpening Remarks / Filecoin: What's Now Possible
38m🛠 Storage & Retrieval on Filecoin Part 1
15mThe State of Filecoin
33mThe FVM Imaginarium Part 2: Early Builders on FVM
14mFilecoin - Building with the Filecoin Onchain Cloud
22mThe FVM Imaginarium Part 1: Request for Startups
31m🛠 Building Applications with Filecoin & FEVM