Mikhail Romanov

Mikhail Romanov, PhD, is a deep learning researcher and engineer. His experience includes deep learning for production, scientific computing and research, accompanied by teaching mathematics and machine learning in general.

His academic experience includes teaching courses at MIPT, HSE, Harbour Space Universities and online platforms. As a researcher, he has conducted research at the Technical University of Denmark, Mail.ru, Samsung Research, Quantori, and Yandex. In his research, his main areas of interest are depth estimation, optical flow, optimisation of neural networks, multi-task learning, self-supervised learning, LLMs and diffusion models. He has published papers on tomography, deep learning, scientific computing, computer vision, generative AI, and diffusion models.

As an engineer, he has worked on products at Tinkoff Bank (natural language processing and data mining), Mail.Ru, Samsung Research (computer vision), ACQuilio (Reinforcement Learning and Recommender Systems), Quantori (Machine Learning), and Yandex (Diffusion Models) with team leading roles at Samsung Research, ACQuilio and Quantori.

As a teacher, he started as a school math teacher (2008-2012), then was TA at the Technical University of Denmark from 2012-2015 during his PhD project. Since 2016, he has developed his courses, one of which has been published in Stepik.Org. This course is taught at MSU, MIPT, UrFU, SFU, and other universities and has approximately 90k subscribers. Since 2019, he has been teaching machine learning and deep learning at MIPT, the Academic Director of Middle DS School (startup), has built 4 online courses on various production topics and conducted approximately 15 academic modules in Harbour Space University mainly specializing in higher math.