Summary
Alexander Panfilov is a machine learning researcher and PhD student based in Tübingen with eight years of experience spanning academic research and industry data science. He works at the intersection of robust representation learning and object-centric models, with an ICLR 2024 oral paper on regularization for combinatorial generalization and contributions to benchmarks like ShiftHappens. Alexander has applied ML in production-like settings (X5, Yandex) and in engineering optimization (Bosch), and his recent ARENA capstone explored adversarially trained LLMs and deception probes. Supervised by researchers at MPI-IS and EPFL, he combines strong empirical skills with a track record of competitive mechanistic interpretability work. He is comfortable moving between applied A/B experimentation, tooling for learning analytics, and cutting-edge theoretical problems in robustness and generalization. Non-obvious: he repeatedly bridges applied industry projects and top-tier research outputs, making him effective at translating real-world problems into publishable ML advances.
7 years of coding experience
4 years of employment as a software developer
Doctor of Philosophy - PhD Machine Learning, Doctor of Philosophy - PhD Machine Learning at University of Tübingen
Exchange (non-degree) Software Engineering, Exchange (non-degree) Software Engineering at Syddansk Universitet - University of Southern Denmark
Bachelor's degree Software Engineering, Bachelor's degree Software Engineering at ITMO University
National University of Science and Technology "MISIS" (Moscow Institute of Steel and Alloys)
English, German, Russian