A data analyst with a J.D., focused on political and civic data in Washington, DC. Below is a portfolio of independent case studies showing how I work through a problem, from public data to a clear, defensible finding.
I'm a J.D. and M.S. Data Analytics candidate (concentrating in AI/Machine Learning) with a background in business operations and legal analysis. I work at the intersection of technical rigor and public policy, using data to identify problems, uncover insights, and support better decisions.
That focus led me to found Giving Voice to Data, where I do political research, survey design, and data-strategy training for campaigns, organizations, and advocates. The case studies below are independent work built to demonstrate that same process in public view.
Five projects built on public data about Washington, DC: who's running, who's turning out, who isn't, and why it matters for policy and outreach.
Built a nonpartisan website comparing candidates in the June 2026 DC primary on their public positions and platforms, translating public statements and voting records into a clear, side-by-side format so voters could compare candidates without digging through campaign sites one at a time.
Joined certified 2026 primary turnout results by ward to Census demographic and economic data to see whether turnout gaps tracked socioeconomic lines. They did, almost exactly.
Ward 8, the city's lowest-income ward with the highest poverty rate and lowest degree attainment, turned out at roughly half the rate of Ward 4, the kind of gap a campaign or advocacy group needs to see before deciding where to invest outreach.
Analyzed the U.S. Census Bureau's Current Population Survey Voting Supplement, the federal government's authoritative survey of voting behavior, to see why registered voters nationally didn't cast a ballot in 2024.
Turnout climbed steadily with education, from 52.5% among high school graduates to 82.5% among those with an advanced degree, the same pattern found at the ward level in Case 02. A caveat worth flagging: CPS self-reported turnout is known to overstate actual turnout, especially in DC, so this data should be read as directional, not exact.
Calculated the share of renters paying 30%+ of income toward rent, HUD's standard threshold for "cost-burdened," by ward. The finding cuts against the obvious assumption.
Ward 8 has the lowest median rent in the city but the highest share of cost-burdened renters, because even cheap rent consumes an outsized share of a much smaller income. Ward 3, with the highest rent in the city, has a comparatively lower burden rate because incomes there can absorb it.
Pulled monthly, DC-specific federal employment data from the Bureau of Labor Statistics to measure the impact of 2025's federal workforce reductions directly on the District, separate from the broader DC-Maryland-Virginia metro numbers most coverage cites.
The decline wasn't gradual: it was concentrated in two sharp drops, one in August 2025 as deferred resignations took effect, and a larger one in October 2025 tied to formal reductions in force.
Sourcing from government APIs, open data portals, and public microdata, joining across sources, and handling missing or inconsistent records.
Identifying patterns, calculating rates and burden metrics, and testing whether a relationship actually holds before presenting it as a finding.
Building charts that make the argument, not just display the numbers, paired with plain-language explanations of what a finding means.
See my full experience below, or reach out about fellowship and internship opportunities.