About

Two B.S. degrees from Virginia Tech in 2004, both with honors — Physics and Applied Computational Mathematics. The decision to do both was straightforward: I wanted to work on problems that involved complex system dynamics, which required the formal mathematics under the physics, not just the physics. By the time I graduated, it was already clear that I was not going to be a physicist. What the training gave me, and what I kept, was the willingness to bring quantitative tools into messy social problems and the patience for work that takes a decade to come into focus.

I came to the University of Chicago for a Master’s in 2005 to find a discipline that took social systems seriously as objects of empirical study — one with theory, methods, and a substantive record of findings about how institutions actually behave when the people running them are not being watched. Sociology had that record. It also gave me something the technical training had skipped: the recognition that capability deployed without careful attention to how social systems actually work tends to produce new harms, usually for whoever was already worst positioned. That recognition has shaped every decision since — what to study, how to study it, who to do it with.

My dissertation and subsequent postdoctoral work at Brown University turned that combination toward urban history. Working with John Logan and Reynolds Farley through Brown’s Population Studies and Training Center, I helped build the geocoding infrastructure that assigned spatial coordinates to full-count census records for 69 U.S. cities, spanning 1900 to 1940 — roughly 30 million records for 1930 alone. That project, published as Logan, Graziul & Frey (2018), was fundamentally a data infrastructure problem at scale: designing ETL workflows that could handle historical inconsistency, integrating legacy code, and developing integrity metrics to know when the process was trustworthy. It was also the first time I confronted a pattern that has defined every major project since: the tools we have were not built for the questions we’re actually asking.

In 2017, during Hurricane Harvey, a coordination problem was unfolding in real time on social media. People in floodwater were posting their locations. Volunteers who could help couldn’t find them. The information existed; the infrastructure to use it didn’t. Someone else put out a tweet that named the problem, and within 48 hours hundreds of pseudonymous volunteers had self-organized into what became CrowdRescue HQ. I joined as one of them. I started by manually geocoding the difficult-to-resolve addresses that automated tools couldn’t place — work where someone had to read a distress message, parse it against local landmarks, and produce a usable coordinate. From there I led the GIS team. Eventually I was asked into the small group that helped the effort transition into CEDR Digital Corps, where I currently serve as Co-Chair of the Board of Directors. What I learned across that arc was not primarily about disaster response. It was about what makes self-organized work durable: the unglamorous internal coordination — trust calibration, leadership legitimacy, socioemotional repair when urgency fades — without which most self-organized efforts dissolve before producing anything that lasts. That is the proof of concept for everything that followed.

Beginning in 2018 at the University of Chicago, I led the first systematic research program treating police radio communications as a dataset for understanding police behavior. The data was legally public — broadcast on radio frequencies anyone can receive. But no governance framework existed for research use of material that was public in legal status while being privacy-sensitive in practice. I built one. Working collaboratively with the UChicago Social and Behavioral Sciences IRB, the University Research Administration, and the Office of Legal Counsel, my team developed data use agreements, a DUA workflow, and a structured repository that allowed rigorous research while managing the genuine privacy and civil liberties concerns the data raised.

The current work runs in two parallel structures. I serve as Executive Director of the Illinois Data Equity Project — IDEP — which develops community-led infrastructure for responsible data governance built around the communities most affected by data-driven systems. I am also the Principal of Graziul Advisory, an AI governance consultancy for organizations navigating the gap between what the technology can do and what law and ethics currently require.

Education

  • PhD, Sociology — University of Chicago (2016)
  • MA, Social Sciences — University of Chicago (2005)
  • BS, Physics & BS, Applied Computational Mathematics (both with honors) — Virginia Tech (2004)

Find me elsewhere