Maksym Tretiakov is a visitor at the ELPIS Lab, where he works on structured stochastic variational inference for Gaussian Process Latent Variable Models (GP-LVMs). This project is supervised by Professor Vincent Fortuin, Dr. James Odgers, and Professor David Rügamer.

His research explores structured variational approximations that capture dependencies in the posterior and improve uncertainty estimation compared with standard mean-field approaches. Maksym is also a Ph.D. candidate in Statistics at LMU Munich.

Previously, Maksym completed his Master’s thesis under the supervision of Professor Vincent Fortuin. His thesis focused on scalable marginal likelihood estimation for probabilistic neural networks using Laplace approximations, particularly in limited-data settings.