My research interests focus on probabilistic machine learning approaches in which the exploitation of structural properties (e.g., symmetries or conditional independencies) enables (exact and) efficient computations. Moreover, I am intrigued by prior specification and approximate inference in Bayesian deep learning, and approximate inference in Bayesian nonparametrics (Gaussian processes and priors on random partitions).

In addition to my research, I’m part of the open-source project Turing.jl, which aims to provide a robust general-purpose probabilistic programming language for all branches of science.