I am a fifth-year Ph.D. candidate in Operations Management at UCLA Anderson School of Management. I received my B.Sc. in Industrial Engineering from Sharif University of Technology.

My research is motivated by a fundamental question: how can we operate critical infrastructure systems more safely and fairly under uncertainty? Much of my work focuses on wildfire risk mitigation in power grids, where poor operational decisions can have severe environmental and societal consequences.

I develop large-scale optimization and learning-based methods for stochastic decision-making problems, with particular interest in reinforcement learning and settings with limited or noisy feedback. Across projects, I aim to bridge theory and practice by designing algorithms that are both analytically grounded and operationally relevant.

You can find my CV here: Download CV (PDF).

• Mar 27, 2026 — Selected to participate in the Rising Stars Workshop at the University of Michigan (29 selected out of 132 applicants).
• Mar 23-27, 2026 — I will present at the 4th International Smoke Symposium (Tallahassee, FL).
• May 7–11, 2026 — I will present at the 36th Annual POMS Conference (Reno, NV).

• Preventing Catastrophic Wildfires: Annual Grid Inspection and Maintenance Plan

  Under review at Operations Research

  [SSRN]

  with A. Fattahi, S. Dasu, R. Ahmadi

  We design an optimal inspection and maintenance plan for large power grids under uncertainty, capturing correlated wildfire risk and operational capacity constraints. Using data from one of California's largest utility companies, we show how principled planning can substantially reduce the damage from catastrophic wildfires.

  • Media coverage: UCLA Anderson Review
  • Finalist, POMS College of SCM Best Student Paper Award 2026
  • Accepted for Early-Career Sustainable Operations Workshop (Flash Talk), 2026; Best Flash Talk Award
• Fair and Flexible Scheduling for Dynamic Energy Load Control under Stochastic Demand

  Under review at Management Science

  [SSRN]

  with A. Fattahi, S. Dasu, R. Ahmadi

  We formulate a stochastic dynamic program for direct load control that jointly models fairness over time and operational flexibility in admissible call lengths. We develop a scalable aggregation–disaggregation method with provable guarantees and quantify the efficiency impacts of fairness and flexibility using high-resolution CAISO demand data.

• Learning to Prevent Utility-Caused Wildfires through Sequential Inspection and Maintenance

  working paper

  with S. Dasu, R. Ahmadi

  We study sequential inspection and maintenance when ignition risk is learned from noisy inspection signals through a shared statistical model. We propose a pessimistic index policy that targets inspections and maintenance using uncertainty-aware risk estimates, and evaluate its performance against natural baselines.

The best way to reach me is by email: abolfazl.taghavi.phd@anderson.ucla.edu.