Personal Info

Hi, I am Martin Wertich, a second-year MSc student in Computer Science at ETH Zurich.

My interests lie in the field of Theoretical Machine Learning, with the focus on the mathematical intuition behind it. I am particularly interested in the mathematics of Learning Theory, Data Science, and large randomized systems such as Random Matrix Theory or Random Graphs. My goal is to pursue a PhD in Theoretical Machine Learning after completing my Master’s degree, to make a small contribution to understanding the inner workings of advanced ML methods, even for non-mathematicians.

I worked as a Research Assistant at the ETH AI Center under the supervision of Barna Paztor, Ido Hakimi and Prof. Andreas Krause from the Learning and Adaptive Systems Group until January 2026, where my student colleagues and I developed an RLHF (Reinforcement Learning from Human Feedback) pipeline for LLMs (Large Language Models) using Active Learning. We submitted our paper to ICML 2026 (International Conference on Machine Learning), where it is currently under review. However, you can already take a look at the arXiv version ActiveUltraFeedback: Efficient Preference Data Generation using Active Learning. Currently, I work on a semester project under Julia Kostin and Prof. Fanny Yang on trying to find and prove new guarantees for Compositional Generalization under noise for additive energy distributions.

Research Interests

The field of Machine Learning Theory fascinates me the most in Computer Science, as it is one of the few CS fields that combines both mathematical rigor and practical relevance for the future. The advancement of new technologies, particularly in Generative Modelling, surpasses our understanding by an increasingly wide margin, and we need to catch up with the forefront of Machine Learning Research.

I find it particularly exciting to puzzle over mathematical proofs, even if it is often initially frustrating and takes a tremendous amount of time (and pain:)), because every solved or unsolved proof advances me and prepares me for the next potentially more difficult one. I am particularly impressed by the theoretical CS research groups (Statistical Machine Learning Group, Decision Intelligence Group, Institute for Operations Research , Learning and Adaptive Systems Lab) at ETH, where doctoral students advance their research with difficult novel proofs.

Short Biography

I obtained my Bachelor of Science degree at Johannes Gutenberg University in Mainz, where I quickly realized that the mathematical subjects, particularly machine learning and mathematical modeling, best matched my interests, in contrast to the more applied subjects. In my bachelor’s thesis, I analyzed the mathematics of Attention Layers in Transformer models.

Fortunately, I had the opportunity to join ETH as a Master’s student, ranked as the best university in Continental Europe for Computer Science. ETH excels in Computer Science and Mathematics, equipping its students with a solid foundation to make significant advances in these fields.

I enjoy teaching students and assisting professors, which I was able to do at ETH as a teaching assistant in Stochastics and Machine Learning (D-MAVT Bachelor) and as a research assistant in Introduction to Machine Learning, where I work on improving Ethel.

Nonetheless, I also worked outside the academic bubble during my bachelor’s degree. I started as an ML engineer, working as a Werkstudent at Schott in Mainz, where I primarily investigated Explainable AI for large time-series data streams. After that, I worked as an intern at Envision Entertainment in Ingelheim as a software developer on the strategy computer game “Pioneers of Pagonia”, which was already released in Early Access, but is yet to be fully released.

Hobbies & Interests

I love the Swiss Alps, and if ETH permits time, I would like to explore them throughout the country. I go hiking and climbing in the summer, while skiing and cross-country skiing during winter. Moreover, I am a fan of board and card games and enjoy playing them to socialize during a game night with my ETH colleagues.