Ryan Li is Surf AI’s cofounder and CEO. He builds AI tools for crypto research and has written a detailed account of using parallel coding agents to optimize a strategy in a simulated prediction-market challenge.[1][3][4]
Start with AI Autoresearch Strategy Generator →
X · ryanli Career
Li’s first-person career account describes interrupting Berkeley study to build an early company, returning to finish, working on Instagram caching infrastructure and later leading a livestreaming engineering team. Surf AI combines model-based research with crypto-specific information sources. The official team page names Li as cofounder and CEO alongside the other founders.[5][1]
Contributions
Li’s challenge report describes agents exploring independent strategy hypotheses and parameter sweeps. He saved findings and dead ends so later agents could reuse them, and occasionally restarted from the rules to escape a local optimum. Evaluation distinguished local seed batches, a fixed-seed leaderboard and fresh-seed final runs. The strategy adjusted quote sizes to estimated arbitrage risk and changed behavior across simulated regimes. Its lesson concerns experimental search and robustness inside that simulator, not a promise of returns in a real market.[4]
Tools, writing & talks
Writing, research & talks
Start here
A first-person account of parallel hypothesis search, parameter sweeps, saved findings and multi-seed evaluation.
Start with the three evaluation layers and the change from one seed batch to multiple batches. Python and basic simulation knowledge help; compare robustness across seeds before drawing conclusions from a leaderboard.
Author and challenge participant[4]Original source