About me

I am a M. Sc. Student in the Department of Statistics at the University of British Columbia, supervised by Ben Bloem-Reddy. My current methodological interests are in latent variable modeling, causal inference, and intersections with machine learning. On the applied side, I have worked in reliability analyses for structural engineering, as well as on a variety of projects via the Applied Statistics and Data Science Group in our department.

Highlights

  • As of September 2022, I will be a Ph. D. student. Supervisory and departmental parameters are unchanged.
  • Our preprint on indeterminacy in generative models is live: arXiv: 2206.00801. We develop a unifying theory of “latent variable identifiability” in generative models. Amongst other things, our work suggests design choices that result in strongly identifiable (i.e., uniquely recoverable) latent variables. If anyone has an interest in applying strongly identifiable latent variable models either for application or new methodology and would like to collaborate, do reach out!

Research

Google scholar does an excellent job at indexing my archived works. This section is hence reserved for additional resources and supplements.

  • poster and paper for Multiple Environments Can Reduce Indeterminacy in VAEs. NeurIPS Workshop on Causal Inference & Machine Learning: Why now? (WHY-21). Note arXiv: 2206.00801 supercedes this workshop paper.