MM-DML
A robust version of double machine learning designed to handle outliers, non-Gaussian residuals, measurement error, and other messy-data problems.
Harvard Government • Time Series • Causal Inference • Interpretable Machine Learning
Causal inference and data science for hard-to-measure, noisy observational data.
My work focuses on building and applying data science methods for messy real-world settings, especially where measurement is difficult and the data are noisy, incomplete, or spatially structured. I am interested in combining machine learning with interpretable modeling so that prediction and substantive understanding remain connected.
- Time series methods
- Causal inference
- Machine learning
- Interpretable data science for complex observational data
Projects
Methods, teaching materials, and applied research builds.
A Basic Introduction to Stats Concepts
A Bookdown-style introduction to statistics and data science built from teaching undergraduates at Harvard.
Using Satellites to Estimate Population
Population estimation with VIIRS night lights at a 10km by 10km grid-cell level.
RobustCause
A library for robust regression, causal inference, and marketing mix modeling.
Research
Selected publications and current lines of inquiry.
When BLUE is not best: non-normal errors and the linear model
A methods paper on robustness, outliers, and why standard least-squares workflows often need stronger diagnostics and estimators.
Albedo as a Measure of Climate Change for Civil War Research
Argues that albedo can capture land carrying-capacity change in ways that are more politically meaningful than coarse climate indicators alone.
Climate Change, Desertification, and Migration
Studies drought, desertification, and urban migration in Syria using albedo and night lights as linked environmental and demographic signals.
Bringing the Civil Back into Civil Military Relations
Reframes coup risk around civilian state and party strength rather than military variables alone.
Teaching and experience
Work that sits between substantive research, pedagogy, and policy relevance.
My teaching and research support roles center on making quantitative methods legible and useful, whether for undergraduates learning data science or for academic leaders shaping research agendas.
Research Fellow
Weatherhead Center for International Affairs, Harvard University
Conducted research for the Center’s leadership, translating complex international relations and political science questions into clear analytic outputs for academic and policy audiences.
Teaching Fellow and Instructional Fellow
Gov 50 Data, Harvard University
Designed and taught core data science material, helping students move from hypothesis formation to data collection, analysis, and public-facing communication.
Teaching Fellow
Behavioural Insights and Public Policy: Nudging for Good
Guided students through applied behavioral economics, intervention design, and evidence-based policy evaluation.
Research agenda
Building interpretable evidence for high-stakes political questions.
The through-line across my work is methodological clarity: using stronger data and better models to explain violence, migration, and state instability in ways that remain substantively meaningful.
Astrophotography
A separate practice in observation, patience, and image processing.
Outside political science, I also capture and process astrophotography. The gallery stays on the site, but as a distinct visual register rather than part of the academic core.
