I am a 5th year PhD candidate in Statistics at the University of British Columbia. I completed a MSc from the same department in 2022. Before moving to Vancouver, I completed a BSc in Statistics at the University of Ottawa, where I am originally from.
My research has primarily focused on causal and probabilistic ML. I have worked on characterizing the identifiability problem in latent variable generative models, which has implications for the identification of counterfactual outcomes under structural causal models. My latest work uses score matching to estimate a new class of structural causal models without making parametric assumptions on the noise distribution, which can be applied to functional causal discovery.
Recently, my interests have shifted toward statistical methodology while continuing to draw on tools and ideas from ML. In particular, I am interested in how tools developed for ML tasks, e.g., sampling and generation, can be repurposed for semiparametric inference and for studying distributional effects. My ongoing work develops estimators for partial effects of continuous treatments along these themes, targeting distributional and other infinite-dimensional estimands using kernel methods, with statistical guarantees.
Outside of UBC, I also worked on causal representation learning and multimodal learning during an internship at Valence Labs/Recursion Pharmaceuticals with Jason Hartford. I also visited the Gatsby Unit at UCL hosted by Peter Orbanz, where I became interested in kernel methods and their application to causal inference.
Alongside my research, I actively work as a statistical consultant both within and outside the department. Most recently, my consulting work has introduced me to methodological problems in clinical trials, including covariate adjustment and group sequential designs. Looking forward, I am interested in developing a more applied component to my academic research, particularly through collaborations in clinical and biomedical applications.