Dissertation · Collaborating Agents & Network Interdiction
How multi-agency collaboration improves labor trafficking interdiction outcomes
Overview
My dissertation develops network interdiction models to understand and improve how law enforcement agencies, NGOs, and government entities collaborate to disrupt labor trafficking in U.S. agricultural supply chains.
Advisor: Prof. Kayse Lee Maass, Operations Research & Social Justice Lab, Northeastern University
Committee: Dr. Shahin Shahrampour · Dr. Gabriela Gongora Svartzman
Expected completion: 2026
Motivation
Anti-trafficking organizations rarely operate in isolation — yet the operations research literature has largely modeled interdiction as a single-agent problem. In practice, Homeland Security investigators, labor departments, NGOs, and local law enforcement all play roles in detecting and disrupting labor trafficking, particularly in agriculture where H-2A guest worker programs create structural vulnerabilities.
What happens when these agencies share information, coordinate interdiction resources, or divide enforcement territory? When does collaboration help, and when does it create unintended consequences? These are the central questions of my dissertation.
Research Threads
Thread 1 — Collaboration in Antitrafficking Efforts: A Network Interdiction Problem
Models how two cooperating interdictors — sharing resources or information — compare to independent enforcement. Shows that collaboration substantially improves detection outcomes under realistic network conditions.
Presented: INFORMS 2025 (Atlanta) · Working paper with K.L. Maass
Thread 2 — Scoping Review of Collaboration and Multiple Attacks in Network Interdiction
A systematic review of the NIP literature on multi-agent and multi-attack settings, identifying gaps and framing the dissertation’s contributions within the field.
Working paper with K.L. Maass
Thread 3 — Unintended Consequences in Network Interdiction: An Agricultural Labor Trafficking Problem
Examines how enforcement strategies that appear optimal in a single-agent setting can produce counterproductive outcomes (displacement, evasion, harm to victims) in multi-stakeholder contexts.
Working paper with Miller, F. and K.L. Maass
Thread 4 — Network Interdiction to Improve Labor Trafficking Detection in U.S. Agriculture
Develops interdiction strategies grounded in the H-2A violation data modeled in our PLOS ONE paper (Jafari et al., 2024).
Under review at Decision Sciences · With Jafari, Bhimani, Farrell & Maass
Methods
- Network interdiction (bilevel programming, mixed-integer programming)
- Multi-agent optimization (cooperative and non-cooperative settings)
- Python with Gurobi for computational experiments
- Public datasets: H-2A violation records, DOL enforcement data (500,000+ records)
- Collaboration with DHS–CINA and agricultural labor enforcement practitioners
Selected Presentations
- INFORMS Annual Meeting 2025 — Atlanta, GA (podium talk)
- INFORMS Annual Meeting 2024 — Seattle, WA
- INFORMS Annual Meeting 2023
- INFORMS Annual Meeting 2022
- RISE Research Expo, Northeastern University 2024 (poster)
- MIE Research Expo, Northeastern University 2023 (poster)