Hello, I'm Kirill Fedyanin.
Hope you are doing good, dear stranger.
I'm currently an ML research engineer, training models for self-driving cars and interested in anything related to physical AI. I would also love to randomly chat about art, self-organizing alife forms, rogue traveling, improvisation plays or really anything else.
If any of it rings a bell, you can reach me at kirill.fedyanin[at]gmail.com.
Below is some of my not very organized work.
My core professional interest now is working on machine learning for the physical world. I'm part of a team working on self-driving cars with dozens of deployed vehicles covering thousands of kilometers daily. I am particularly charmed by decision-making intelligence, and my effort is focused on reinforcement learning for decision planning and execution.
I am also fascinated by the growing industry of humanoid robots -- and all the variations of able-to-work-in-human-context robots. I feel like it would allow us humans to spend more time on our social, artistic and spiritual parts. I also just think it's a very interesting problem to solve.
I've designed some fun problems for the finals of the International Olympiad in AI . Check the problem from the 2026 edition. If you can score 50+, you'll be better than many live competitors.
My first memorable problem was about forcing diffusion network to always draw fire hydrants with a cow
