about me

I’m a Model Risk Officer at Capital One with a background that probably looks unusual for this kind of role. I did my PhD in integrative biology at UT Austin, studying how animals make decisions: mate choice, foraging, social group selection. The core question I kept returning to was: when does a model of behavior make accurate predictions, and when does it quietly fail? That question turns out to be more useful in finance than I expected.

After grad school, I spent several years on the first-line side of ML: building models, scaling infrastructure, and figuring out what it actually takes to keep a model behaving in production. That work taught me that model failures are usually operational: drift, edge cases, features that degrade quietly in production. But the operational failures almost always trace back to assumptions that were never seriously questioned at design time. The interesting work is connecting those two things.

I moved to model risk because I wanted to work on that gap more deliberately. My focus is on the assumptions embedded in model design and elevating modeling practice across the company.

interests

I’m most interested in how things fail, not through obvious breakdowns, but through drift: the slow accumulation of locally rational decisions that erode a system’s margins until something gives. I find this pattern to be quite general.

I’m also drawn to the idea that models—of behavior, of risk, of the world—are useful fictions. Where do these fictions break down? How do we ensure we understand their limitations?