Work

Problem

Standard causal methods assume the relationships through which an effect travels stay fixed. In the settings I study they do not — the intervention itself reshapes the network, and the usual estimators stop meaning what they claim to mean.

What I did

My dissertation develops methods for evaluating an intervention when the structure through which its effects spread is itself changing, and applies them to political networks observed over time.

Result

Effects estimated along a trajectory, not at a pointOngoing doctoral research at Northwestern.

What it demonstrates

Working at the point where an existing method breaks down, and building something defensible rather than applying a familiar tool to a problem it does not fit.

Full research page

Problem

A National Science Foundation project depended on evidence that did not yet exist in usable form. The underlying records were public, but dispersed across sources, formats, and more than a decade of inconsistent reporting.

What I did

Built the data infrastructure end to end: collection, reconciliation of inconsistent records, validation, and the transformation of raw public filings into a structured analytical foundation.

Result

300,000+ observations, spanning over a decadeA reusable dataset now serving as the empirical base for ongoing analysis.

What it demonstrates

Comfort with messy, incomplete, real-world information — and the discipline to build infrastructure other people can rely on, rather than a one-off analysis.

Problem

Launching a consumer coffee business with limited capital, no established brand, and no margin for slow decisions.

What I did

Helped build and operate a consumer coffee business, leading growth strategy, digital marketing, pricing, performance analysis, and the recurring decisions required under limited resources.

Result

From launch to repeatable growthBuilt and operated from zero, under real constraints.

What it demonstrates

Accountability for outcomes, not only analysis. Acting under uncertainty with incomplete information, and living with the consequences of the decision.

Problem

Technical and political ideas lose their substance when they are simplified for general audiences, and lose their audience when they are not.

What I did

Contributed to 49W, a social-science media platform, alongside published analysis in English and Turkish for outlets including Anadolu Agency and Mesail — media production and public writing aimed at holding both rigor and reach.

Result

A community of 380,000+ subscribers reachedPublished analysis in two languages.

What it demonstrates

Translating complexity without losing the substance.

Problem

Statistics is where many social science students decide they are not quantitative people. Reversing that means meeting very different levels of technical background in the same room.

What I did

Taught empirical and statistical research methods at Northwestern, supervised research teams, and coordinated the Comparative Politics Workshop across an academic year.

Result

150+ students taught · 5.4 / 6 evaluationAcross empirical and statistical research methods.

What it demonstrates

Structuring an explanation for the audience in front of you, and leading people rather than only analysis.