OR for Social Good · Prior Research
Bayesian optimization for border security · DEA for women's crisis center networks
Overview
Before my dissertation, I applied operations research to two distinct social problems across two countries — each involving real partnerships with government agencies and each producing findings that directly informed policy.
Project 1 — Optimization of Port-of-Entry Operations with Anti-Human Trafficking Focus
Institution: University of Texas Rio Grande Valley (MS Thesis, 2019–2021)
Advisor: Prof. Hiram Moya
U.S. ports of entry process millions of crossings per year, yet agents have limited time and information to identify human trafficking victims among travelers. This project engineered a Bayesian optimization decision support tool that integrated cost-benefit analysis with statistical predictors to improve the accuracy of trafficking identification at border checkpoints.
The model was developed in partnership with DHS–CINA (Center for Countering Human Trafficking) and published through the Decision Sciences Institute (2020) and as a full MS thesis (2021).
Key methods: Bayesian optimization · Cost-benefit analysis · Statistical modeling
Partner: Department of Homeland Security – CINA
(Drummond, 2021; Drummond & Moya, 2020)
Project 2 — Relative Efficiency Analysis of the Women’s Crisis Center Network in Rio Grande do Sul
Institution: Universidade Federal do Rio Grande do Sul (MS Thesis, 2017–2019)
Advisor: Prof. Denise Lindstrom
In collaboration with the Secretary of Security of Rio Grande do Sul, this project applied Data Envelopment Analysis (DEA) to evaluate the relative efficiency of women’s crisis centers across 450+ cities in the state — home to over 10 million residents.
The models identified best and worst practices in resource allocation, providing actionable recommendations that directly informed the state’s policy decisions on service distribution and investment.
This project also required building a cross-sector partnership with law enforcement, social services, and government stakeholders — an experience that fundamentally shaped how I approach collaborative research. When stakeholders felt genuinely included in the conversation (not just data sources), the quality of the work and the uptake of findings both improved dramatically.
Key methods: Data Envelopment Analysis (DEA) · Large-scale data analysis (R, Python) · Policy-relevant modeling
Partner: Secretary of Security, Rio Grande do Sul, Brazil
Impact: Policy recommendations affecting 10+ million residents
Connecting Thread
Both projects share a conviction that operations research is most powerful when it is built in relationship with the people affected by the problem. The Bayesian model for border security would have been incomplete without DHS practitioners explaining how agents make decisions in the field. The DEA model for Brazil would have missed critical contextual factors without the Secretary of Security’s staff explaining how services actually operate.
That lesson — that good models come from listening — is now at the core of both my research and my teaching.