My research focuses on the interface between deep learning and probabilistic modeling. I am particularly keen to develop models that are more interpretable and data efficient, following the Bayesian paradigm. To this end, I am mostly trying to improve deep probabilistic models with better priors and more efficient inference techniques.