Research

Queueing Discovery

Queueing Causal Models: Comparative Analytics in Queueing Systems

Job market paper · Published in Manufacturing & Service Operations Management

With Opher Baron, Dmitry Krass, Arik Senderovich, and Mark van der Laan.

2025, 28(2), pp. 517–536; issue date March–April 2026.

My job market paper uses queueing data structure to build machine learning models for counterfactual policy evaluation. Classical queueing models can be sensitive to assumptions, while black-box learning may overlook queueing dependencies. We use event ordering to construct a sufficient history as candidate inputs, then learn relationships and conditional distributions. Numerical experiments show accuracy comparable to queueing-theoretic benchmarks where analytic solutions are available. Ongoing work extends these ideas to inventory systems.

Counterfactual validity relies on causal and finite-history assumptions.

Inventory extension: Structural Causal Inventory Models: Identification and Policy Coverage.

• Journal
• SSRN

• 🥇 First prize, CORS 2025 Queueing and Applied Probability SIG
• 🏆 Winner, 2024 Oded Berman Student Paper Competition
• 🎖️ Finalist, CORS 2024 Best Student Paper Competition (Open Category)
• Service Management SIG, MSOM 2024
• 2023 Rotman TD MDAL Research Grant

Capacity Control

Managing Scarce MRI Capacity in Overloaded Queueing Systems

Working paper · Major revision at M&SOM, 2026

With Adam Diamant, Andre Cire, Eugene Furman, and Opher Baron.

We use multiclass fluid models and optimal control to allocate scarce MRI capacity across patient classes with different waiting costs. Under the model assumptions, we characterize a sequential-activation policy: the set of classes eligible for capacity expands over time. We also study minimum service-proportion constraints and a soft fairness regularizer. Our case study uses over seven million MRI encounters across ten years. This model-based control foundation complements my research on learning operational dynamics.

• SSRN
• Public PDF

• 🎖️ Finalist, CORS 2026 Healthcare SIG

Online Learning

Learning to Price in Queues: Arrival Observability and Value Randomness

Working paper · Draft available upon request

With Philipp Afèche, Opher Baron, and Dmitry Krass.

We study a learning counterpart of Naor’s strategic queueing framework, where prices affect joining, congestion, and subsequent learning. Core results concern an unknown common deterministic value and observable arrivals. Direct state-adjusted Kleinberg–Leighton pricing can incur linear regret along a sequence of horizons. Admission-adaptive and regenerative algorithms achieve O(log log T) regret under return-time conditions, matching a minimax lower bound for this model class. The broader project examines arrival observability and value randomness.

• Request draft

• 2024 Rotman TD MDAL Research Grant

Leveraging Advanced Analytics to Streamline the Emergent Dialysis Process at Parkland Hospital

Published in Healthcare Analytics, 2025

With Olga Bountali, Sila Cetinkaya, Michael Hahsler, Farnaz Nourbakhsh, and Henry Quinones.

Journal version: An investigation of treatment barriers for End-Stage Kidney Disease patients using advanced analytics. 8, article 100438; December 2025.

We combine descriptive analytics and simulation to understand barriers to emergent dialysis and evaluate operational redesign at Parkland Hospital.

• Journal
• SSRN

A High-Fidelity, Machine-Learning Enhanced Queueing Network Simulation Model for Hospital Ultrasound Operations

Published in Winter Simulation Conference, 2021

With Yihan Pan, Jin Guang, Xinyun Chen, J.G. Dai, Chengwenjian Wang, Xuanming Zhang, Jingjing Sun, Pengyi Shi, Yichuan Ding, Song Wu, Kai Yang, and Hongxin Pan.

We build a calibrated queueing-network simulator for a hospital ultrasound center, using machine learning to capture complex patient routing. This project motivated my subsequent research on learning operational dynamics from data.

• IEEE

Structural Causal Inventory Models: Identification and Policy Coverage

Work in progress

With Opher Baron and Dmitry Krass.

We retain known inventory identities, learn stochastic demand and supply mechanisms, and study whether observed trajectories support evaluation of a target ordering policy.

Queueing Causal Models for Emergency Department Efficiency

Work in progress

With Opher Baron, Dmitry Krass, Mark van der Laan, and Yufeng Zheng.

An ongoing application of queueing causal models to emergency-department operations and patient-flow decisions.

• 2025 Rotman TD MDAL Research Grant

Fairness in Multi-Stage Healthcare Access: A Study of Congestion, Chronic Patients, and Operational Redesign

Work in progress

With Olga Bountali, Sila Cetinkaya, Michael Hahsler, Farnaz Nourbakhsh, and Henry Quinones.

We study how repeated deferral in a re-entrant healthcare access system distributes waiting burdens across patient trajectories, and how operational memory and allocation rules change these burdens.

• CORS 2026 Annual Meeting