I am currently working at Unlearn in San Francisco, CA, as a Machine Learning Engineer, since November 2022. I create generative multivariate time-series models for clinical trial outcomes, which are used to create “Digital Twins” of patients. You can read about details in this tech report we published.

---

I enjoy getting into new domains, finding the problems I can solve and delivering solutions. Most of the time, I find myself building tools, be they software or experimental, to push the envelope of what is possible. I have for a long time had a passion for applying mathematical models to real world problems, be it optical femtosecond laser systems, hard X-ray instrumentation, time series or spectroscopic data. In my current position, I am examining clinical time series, which have a lot of unique challenges. The data is small, but expensive to acquire, rife with missingness, and tabular. So, data efficiency and generalization of performance are what I focus on chiefly. In previous work, for example with scientific data, data was relatively cheap to acquire and enormous (many terabytes), but accuracy of prediction was of utmost concern.