About

Hi! I’m a PhD student in the Evolutionary Intelligence group at the Dutch Institute for Maths & Computer Science (CWI).

My primary research interests are hyperparameter optimization and Neural Architecture Search (NAS). I believe leveraging Evolutionary Algorithms (EAs) for non-differentiable Machine Learning problems such as these is a potentially high-impact research direction. In addition to not requiring gradients, most EAs are trivial to parallelize & scale, so they are fully compatible with Sutton’s Bitter Lesson.

I’m convinced that computer science is a strong-link problem: it does not matter how many weak papers there are, the few strong papers will push the progress forward. I try to conduct my own research accordingly.

Always happy to chat about these topics! Drop me a line at alex@chebykin.dev

Research-wise I’m also interested in explainable AI (my Master thesis was partly about using NAS for explainability) and image generation (the applied part of my PhD project is about privacy-preserving medical image data sharing with GANs). Life-wise I’m into fantasy, hiking, and discussing weird ideas.

CV