Looking for a Research Lead to build with us in SF @trytalp. The dirty secret in AI simulation right now is that almost everyone is selling theater. Clean demos, zero proof. Nobody has published real benchmarks, confidence intervals, or calibration against actual human behavior. The team that turns this from vibes into an actual science wins the category. What I'm looking for: 1) You know what’s actually happening under the hood. You didn't just pick up ML from Twitter threads or high-level APIs. You respect sound statistics, live in PyTorch/JAX, and have real intuition for how models break at the edges. 2) You despise fake progress. If an experiment shows a 5% bump, your first instinct isn’t to celebrate—it’s to hunt down the eval leakage or sampling bias. You’d rather kill your own favorite idea than ship a hallucinated win. 3) You care about production, not paper count. Academic rigor matters to us, but pure ivory-tower research is boring. What counts is whether the system survives contact with messy enterprise data. 4) Extreme autonomy. You don’t wait around for someone to write you a ticket. You look at an ambiguous problem, figure out what question actually matters, and go get the answer. 5) You raise the bar for everyone around you. You’ve led technical work before, you have taste in talent, and you know how to build a culture where sloppy work has nowhere to hide. Hybrid in SF · $200k–$300k + real equity (raised $2.2M, backed by Formus, Sunshine Lake, Aito vc, a16z Scout Fund). If you want to take a real swing at this, shoot me a DM or drop an email to baran@talp.ai with a few things you’ve built or broken.