Research

I work on causal machine learning and educational data science. The four threads below run through my dissertation and my collaborations. Each card opens a short page with the methods, the data, and the papers.

Optimal and dynamic treatment regimes

#Policy Learning #Optimal Treatment Regimes #TMLE #Feasibility

Learning which sequence of courses to recommend to an individual student, and making the recommendation one a school could actually follow.

Generative AI for synthetic data and method evaluation

#Synthetic Data #Monte Carlo #CTGAN #Diffusion Models

Using generative models to build simulation studies whose data behaves like the data researchers actually analyze.

LLM multi-agent systems for automated quantitative research

#Multi-agent Systems #LLM Evaluation #Automated Research #Distillation

EDM-ARS carries a study from question to manuscript, and reviews its own methods on the way. Plus distilling a black-box model into one that explains itself.

Algorithmic fairness in prediction and decision-making

#Algorithmic Fairness #Dropout Prediction #Auditing #Multilevel Data

A model can be accurate overall and wrong unevenly. Measuring that before anyone acts on its output.