AI Research Engineer

You build the system that does our research: agents that form hypotheses, run experiments against markets, and learn from what comes back, plus the models that make them better over time. Right now that means rigorously studying how automated research applies to problems like objective-function design, and building a leaderboard where LLMs run agentic research to develop algorithmic strategies in a production-grade environment. We're working to take ourselves out of the loop, and most of the hard problems are still open. WHAT YOU'D DO - Design the loop itself: evolutionary search in the spirit of AlphaEvolve, and agent orchestration for unbounded, autonomous research - Train and improve the models that drive how the system reasons, evaluates, and decides - Design the objective functions and rewards the agents optimize, so the system chases real progress instead of gaming the metric - Turn what each run learns into knowledge the system compounds on, so it gets sharper over time YOU MIGHT BE A GOOD FIT IF - You have real depth in ML, as a researcher or an engineer (many of us have PhDs, but what you've built matters more than the credential) - You can take a vague, open-ended problem and drive it to a working result on your own - You're strong in Python and have trained models that made it into production - You prefer fast loops and simple experiments to long solo efforts BONUS - Experience with agents and LLM tool-use, reinforcement learning, or time-series modeling - You've worked near markets, or another domain where the feedback is fast and unforgiving DETAILS - Stack: Python, PyTorch, LLM and agent tooling, backtesting infra - Location: San Francisco (in-person) - Comp: $175K to $250K base, plus 0.25% to 1% equity