Research (By Topic)My research focuses on the design and deployment of AI systems that empower human decision-makers and improve societal outcomes. Operating at the intersection of machine learning, operations research, and economics, I investigate the complex dynamics of human-AI interaction. To bridge theory and practice, I develop novel methodologies in contextual bandits, transfer learning, and policy optimization, and rigorously test these approaches through large-scale field deployments in healthcare, education, and public policy. Selected papers in each direction (1) AI for social impact, (2) ML/AI Algorithmic Methods, and (3) human-AI collaboration are included below.
AI for Social ImpactI am especially excited about partnering with governments, NGOs, and schools to study how/where AI can improve human outcomes in the field.
Education and human capital developmentEffective Personalized AI Tutors via LLM-Guided Reinforcement Learning (* equal last authorship) Self-Regulated AI Use Hinders Long-Term Learning
Generative AI Without Guardrails Can Harm Learning: Evidence from High School Mathematics
(* equal first authorship); cited in the 2025 Economic Report of the President
Healthcare and public sectorImproving Access to Essential Medicines via Decision-Aware Machine Learning
(* equal last authorship) 1st Place, Public Sector in Operations Best Paper Award (2024) Efficient and Targeted COVID-19 Border Testing via Reinforcement Learning
(* equal first authorship) 1st Place, Pierskalla Award for Best Paper in Healthcare (2021) Adaptive Clinical Trial Designs with Surrogates: When Should We Bother?
Unmasking Human Trafficking Risk in Commercial Sex Supply Chains with Machine Learning
(collaboration with TellFinder Alliance & Uncharted Software) 1st Place, MSOM Student Paper Competition (Ramchandani, 2022) Bandits for Online Calibration: An Application to Content Moderation on Social Media Platforms
Machine Learning AlgorithmsI build methods that can learn useful data-driven decisions from limited, high-dimensional data, often dynamically or across multiple sources. My focus has been on policy optimization, contextual bandits, and transfer learning.
Policy OptimizationWinner’s Curse Drives False Promises in Data-Driven Decisions: A Case Study in Refugee Matching
Beating the Winner’s Curse via Inference-Aware Policy Optimization
Bandits & Transfer LearningOptimal Multitask Linear Regression and Contextual Bandits under Sparse Heterogeneity
Multitask Learning and Bandits via Robust Statistics
Group-Sparse Matrix Factorization for Transfer Learning of Word Embeddings
Meta Dynamic Pricing: Transfer Learning Across Experiments
Predicting with Proxies: Transfer Learning in High Dimension
Mostly Exploration-Free Algorithms for Contextual Bandits
Online Decision-Making with High-Dimensional Covariates
Human-AI CollaborationI'm interested in how AI systems should be designed to collaborate with (rather than replace) humans. Humans have limited attention, require incentives, may be unfair, and learn over time; these factors shape human-AI collaboration. The Human-AI Contracting Paradox
Action vs. Attention Signals for Human-AI Collaboration: Evidence from Chess 1st Place, Decision Analysis Society Student Paper Award (Poulidis, 2025) Perceptions of Fairness in Algorithmic Decision-Making
Rethinking Algorithmic Fairness for Human-AI Collaboration
Improving Human Sequential Decision-Making with Reinforcement Learning
1st Place, Behavioral OM Best Working Paper Award (2021) Interpretable OR for High-Stakes Decisions: Designing the Greek COVID-19 Testing System
(* equal first authorship) 1st Place, Wagner Prize for Excellence in Operations Research Practice (2021) Interpreting Predictive Models for Human-in-the-Loop Analytics
|