Michael Leon Chrzan

Data Scientist bridging K12 education practice, policy, and advanced analytics through machine learning, causal inference, and NLP to drive educational equity and evidence-based improvement at scale.

May 2026

I’ll be returing to The Farm to present two projects at the Stanford Education Data Science Conference this May. The accepted works - “The Lanugage of Closure” and “Multimodal Speaker ID in Classroom Environments” - are detailed in my Projects section. “The Language of Closure” has been posted as an EdWorkingPaper preprint as well.

March 2026

March has been a big month!

Two New EdWorkingPapers!

I’ve published two new working papers, now available on Annenberg’s site.

In the first paper, “The Language of Closure: Examining Racial Differences in How A Community Discusses School Closure Metrics”, we examine to what extent race related to community members feedback for a large urban school district’s closure metric decision process. Using BERTopic, a neural topic modeling approach, we analyzed 4,159 suggestions from 2,006 community members and identified 14 coherent topics that capture community priorities. We found significant difference in some topics’ prevalence by race. Furthermore, when we examined the topic outliers we found that White respondents were significantly more likely to provide suggestions that fell outside of community-wide themes. We discuss in the paper some implications for our partner district as well as other districts engaging in similar closure work.

The second is the paper I presented at AEFP 2026, “A Sandbox for Hard Choices: Using Simulation to Explore School Closure Scenarios and Their Consequences”. In this paper we worked with a large urban school district to help them optimize their school closure process for both fairness and efficiency and analyzed 7 closure policies across 5,040 simulated scenarios. We found that (1) equity safeguards materially improve proportionality, but they come with lower cost savings and fewer seats removed, (2) common closure criteria such as enrollment and academic performance are weak tools for achieving both efficiency and fairness (3) rank ordering matters a lot, the sequence used to evaluate schools can substantially change the final closure scenario and (4) there existed a composite ordering rule that combines academic, operational, and equity information performs best overall in meeting the district’s goals. The code for the scenario generation algorithm is available in the project repo.

AEFP 2026

I’m excited to present “A Sandbox for Hard Choices: Using Simulation to Explore School Closure Scenarios and Their Consequences” at the Association for Education Finance and Policy (AEFP) conference, a new paper examining a policy sandbox approach for examining school closure policies written with my co-author and advisor Dr. Francis A. Pearman, II at Stanford University.

I’m also honored to be chairing a session at the conference. “LLMs for Qualitative Coding and Measure Development” featuring work from researchers at Brown, Stanford (Go Trees!), UT-San Antonio, and the University of Michigan (Go Blue!).

The session will take place 10:00am-11:30am, Thursday March 19th. Paper titles are below.

  • “Using large language models in qualitative analysis: Do LLMs match human insights into teachers’ career decisions?” by Andrew M Camp, Kate Donohue, John Papay, Emily Qazilbash, Macy Brammer
  • “Moving past human labels: Best practices for annotating conversational data” by Julian Bernado, Ana T Ribeiro, Xander Beberman, Susanna Loeb
  • “Items and Long-Run Outcomes: How to Measure Achievement and What Drives It” by Jonathan Moreno-Medina, Eric Nielsen, Viviana Rodriguez
  • “Measuring the Occupational Content of College: A Multidimensional Framework for Labor Market Alignment” by Linday S Leasor

July 2025

Published working paper: “Deeper Roots Before the Storm: Utilizing Machine Learning to Alert School Districts of Permanent School Closures” with Francis A. Pearman, II and Benjamin W. Domingue (EdWorkingPaper: 25-1210, Annenberg Institute at Brown University).

June 2025

Graduated from Stanford University with M.S. in Education Data Science (GPA: 4.1). Specialized in causal methods, measurement, networks, and natural language processing.

April 2025

Started position as Data Scientist at the Center for Educational Data Science and Innovation (EDSI) at the University of Maryland, College Park. Leading research initiatives on AI for equity in education.

Presenting “Foregrounding equity in school closures, mergers, and co-locations” at the American Educational Research Association (AERA) Annual Meeting in Denver with co-authors Khanna, R., Pearman, F. A., Wentworth, L. P., Kim, M., & Lau-Smith, M.

March 2025

Two of my dear friends from Stanford and I were featured on an EdTech Insiders episode, The Promise and Perils of AI in Education, where we got to discuss our current work in our program. Check it out below!

September 2024

Completed data science contract work with San Francisco Unified School District, developing scenario generation algorithm evaluating 400+ viable school closure scenarios with equity-focused indicators.

August 2023

Awarded Dean’s Fellowship to pursue M.S. in Education Data Science at Stanford University Graduate School of Education.

2022

Received Rookie Chess Coach of the Year award from DPSCD STEM Activities Department for building the chess program at my school.