I research statistical and causal machine learning (ML) at Fero, which I co-founded five years ago. I also teach and advise students at Columbia. I volunteer by serving as production editor of JMLR, and as an area chair for NeurIPS, and ICML.

I am motivated to explore how ML (and, in general, software) interfaces with climate change. To this end, I am teaching a course about Machine Learning and Climate. I have also started a blog to capture some thoughts in this space.

I currently research trustworthy ML methods — algorithms that carefully move beyond correlations, quantify future uncertainties, and identify causal relationships. I find these qualities to be a key component for a broader adoption of ML methods in industry.

I previously worked on approximate Bayesian inference with David Blei and probabilistic programming with Andrew Gelman. I designed Stan's variational inference algorithm. I obtained my Ph.D. at Yale, where my thesis won a best thesis award.

My last name is pronounced “cue-choo-kell-beer.”