Keduse Worku

I’m a final-year astrophysics PhD at Johns Hopkins University (finishing Summer 2026), working at the intersection of statistical modeling, Bayesian inference, and complex systems. Currently, I do AI model evaluation at Anthropic while transitioning into applied data science and ML roles in industry.

Most of my work involves pulling weak signals out of messy, high-dimensional systems. In practice, this has meant building probabilistic models, running large simulations, and wrangling observational datasets where uncertainty actually matters. I’ve applied these tools both to studying the early universe and, more recently, to financial modeling.

I’m most excited by problems where prediction alone isn’t enough. I care about understanding how models behave, where they fail, and what tradeoffs matter.

I also enjoy communicating these insights clearly, whether to fellow researchers, engineers, or anyone curious about the science.