Pixel-art lab scene with a horizontal four-node pipeline (paper, reading-glasses, medical-cross, protein-structure) connected by cyan data lines. Stanley on the right with clipboard, mid-gesture pointing at the pipeline.
DeSci Active

Lilypad Network

Head of Research at Lilypad — decentralized compute marketplace, GPU verification research, and the Altruistic AI Agents series.

decentralized-computegpu-verificationaltruistic-aiarxivoncology
GitHub →

Lilypad Network is a decentralized compute marketplace — Uber for GPUs. It connects GPU owners with compute consumers in a trustless, verifiable marketplace. Researchers shouldn’t need to wait eighteen months for compute time. Lilypad solves that.

As Head of Research, I co-authored “Validation of GPU Computation in Decentralized, Trustless Networks” (arXiv 2501.05374) — addressing the hard problem of verifying that remote GPUs actually ran your computation correctly. Exact GPU recomputation fails due to non-determinism. Trusted execution environments require specialized hardware. Fully homomorphic encryption is prohibitively costly. Our solution: model fingerprinting plus semantic similarity plus GPU profiling, producing a binary reference model with ternary consensus.

I also launched the Altruistic AI Agents series — building AI agents for real-world positive impact. Episode one: a four-node agentic oncology pipeline with a Cedars-Sinai researcher, running paper search, paper reading, oncology reasoning, and protein design on Lilypad’s decentralized compute. Someone very close to me went through cancer and didn’t survive. This work is personal.

Lilypad bridges Lattice Protocol’s trust layer with real clinical use cases — the theoretical foundation for verifiable compute meets the practical need for trustworthy AI in medicine.