The Final Thesis Defenses at Timiryazev Academy Are Over

I decided to write this post to remember my first experience serving on examination boards for state exams and final thesis defenses. Let me say right away: it was interesting and an honor, but also extremely demanding and time-consuming.

Yesterday I heard the final group of master's students in AI at Timiryazev Academy defend their work. Fifteen students presented ten projects. For some reason, several students worked in groups; I still don't quite understand that. We didn't do it that way.

The projects varied widely, but almost all centered on training models to recognize and analyze images or other patterns. To be candid, many of them were essentially assemblies of existing open technologies rather than something truly new. That's not criticism, just an observation. Even so, everyone received an excellent grade because they had demonstrated that they'd mastered the curriculum and could build software.

Among the stronger projects, I'd highlight one developed as part of the Big Livestock Farming hackathon (a mature, practical project for detecting lameness in dairy cows); a system for tracking Timiryazev faculty KPIs (a genuinely operational system used by the academy's instructors); and a project to assess tractor and self-propelled equipment operators' attention using alpha brain waves. The last one was entertaining: the student built a headband prototype with two sensors, on the forehead and the ear, to record alpha-wave activity while driving. It was an ambitious direction, though I didn't see much practical value beyond learning to capture neural signals. Still, impressive.

One student's thesis cited more than 100 references, all in English. On impulse, I decided to tease him a little and asked, in English, why he hadn't used any Russian-language sources. The committee members looked surprised, and the students laughed. =) Sorry—I wasn't trying to show off; it was just an impromptu joke. His project looked strong at first glance, although I never fully understood what his AI agents actually produced. Another point worth mentioning: many students overcomplicated their presentations. They described straightforward problems in academically polished language that was almost impossible to follow unless you were already an expert in the subject. Keep it simple. Write clearly. Explain the problem with examples so that people can understand it.

By the way, if any of the students whose theses I reviewed want to discuss their work, find an internship, or look for a job, contact me through the form on my website. I'm open to different kinds of collaboration; who knows, maybe we'll end up working together. Also, I ran out of books for some students yesterday. Message me if you'd like one—I'll gladly give you a copy.

In summary, it was a great experience. I'm glad I tried it, and I believe I gave it my full effort, acted honestly, and graded fairly. I enjoyed the events, but I'm not sure I'm ready to devote this much time to them again—especially if I don't see real engagement from the students, ideally in the form of a like on this post. ;)