Lilypad Network: Serverless Distributed Compute for AI
Listen Now
About This Episode
In this episode of DevNTell, Narb welcomes Alison Haire and Phil Billingsby from Lilypad Network to discuss their serverless and permissionless distributed compute platform. Developed initially at Protocol Labs, Lilypad democratizes access to high-performance computing, especially for AI applications. Alison and Phil provide a deep dive into Lilypad's vision, technical implementation, and the exciting future with its incentivized testnet on the horizon. The episode showcases how Lilypad makes running complex AI tasks as simple as an API call while upholding the decentralization ethos of Web3.
Key Takeaways
Lilypad Network democratizes access to high-performance GPUs and CPUs by creating a decentralized three-sided marketplace for compute jobs.
The platform uses a job-based system rather than a time-based one, making it as easy as an API call to run complex AI inference or fine-tuning jobs.
Lilypad is EVM-compatible and uses an ERC20 token for payment, leveraging Arbitrum for verification and validation of off-chain computation.
Trust and verification in a permissionless system are maintained through on-chain guarantees and game-theoretic methods like optimistic reproducibility.
The network supports a range of models including Stable Diffusion, Llama, and even scientific models like AlphaFold 2, allowing for diverse compute tasks.
Featured Guests
Alison Haire
CEO @ Lilypad Network
Phil Billingsby
DevRel Engineer @ Lilypad Network
Listen On
Share This Episode
Share on XWatch Episodes Live!
Subscribe to our event calendar and never miss a live episode.
View Event Calendar