About me

I am an AI architect / ML engineer working on applications of large language models. My formal educational background is in physics, engineering, and statistics and my past research interests focused on scalable graphical models for large spatiotemporal data.

How did I get here?

I greatly disliked math & physics as a teenager until experiencing excellent teaching and advising from the faculty of the physics department at Macalester College which completely reoriented my life. Around this time, multiple floods struck my hometown of Valley City, ND and I entered grad school at Duke focusing on hydrology and forecasting.

Ultimately, I found that too much of that problem involved uncertainty that wasn’t well-handled by deterministic methods. Around this time, I also got the chance to dip my toes into graduate work in statistics, being enticed to learn more by courses like STA-663 and STA561. I ended up doing the coursework for both the PhD in engineering and the MS in statistics concurrently.

There were only two innovations in my doctoral work. The first was the creation of a scale-invariant formal framework for identifying bodies of water subject to unnatural controls on their influx and outflux. The second was the development of a differentiable hydrology model using Theano circa 2016 to enable the use of Hamiltonian Monte Carlo and later on, gradient descent for variational Bayes. I did a postdoctoral stint at Oak Ridge National Laboratory and transitioned into ML engineering as a career in 2022.