Anton Polishko, PhDFounder

United States - San Francisco Bay Area

I build Principle, a simulation engine that stress-tests business decisions before you commit capital. We create digital twins of companies, competitors, regulators and market forces, then run hundreds of scenarios using techniques from military wargaming.
Before Principle, I was at Google/YouTube building simulation infrastructure for recommender systems. I led the effort scaling an agent platform that let Google Brain researchers run live experiments on YouTube's infrastructure. That work became a SIGIR 2024 publication on user simulation.
Turns out the same techniques that simulate user behavior at YouTube scale can simulate how markets, competitors and regulators behave. That's what we're building now.
PhD in Computer Science from UC Riverside. Published research on Ukrainian language models (LREC-COLING 2024) recognized in GIJN's best data journalism picks of 2025.
At Principle I lead AI infrastructure, simulation architecture and model development.