Armen A. Agalyan

About

I earned my master's in Applied Mathematics from Lomonosov State University, where I worked on Extreme Value Theory in the Laboratory of Probability.1 After graduating, I moved into industry and began working in machine learning on large-scale consumer data systems. For the past five years, I have taken end-to-end ownership of products, developing and deploying models and pipelines that deliver measurable impact across diverse ML domains.

As of recent years, I've shifted much of my focus toward applied AI and generative models. Last summer, I became interested in mechanistic interpretability and was inspired by the Anthropic blog posts. Next, I plan to return to academia and pursue research in the AI safety domain. I'm especially interested in helping close that gap between research prototypes and reliable products.

Outside of work, I enjoy stepping into new areas and challenges. A year ago, I collaborated with the online school PASV as a co-author and lecturer for a machine learning course, where I helped guide more than 500 students through topics including calculus, linear algebra, probability, and statistics. I developed materials for more than 50 lessons, created accompanying homework assignments, and conducted over 200 hours of live lectures and seminars.

Armen Agalyan

now

Last updated: April 2026

Based in Chicago. Consulting on ML systems through Magical Labs and doing independent research on the side.


my favorite things

Math, skiing, making friends, and the Tool.


favorite records

Upcoming...


contact

Best reached by email.


  1. Research focused on the rate of convergence in limit theorems for maxima of random variables in extreme value theory.