About Me

I'm a postdoc in the Center for Systems Science and Engineering at Johns Hopkins University, where I also received my PhD. My primary research focus is developing AI and optimization models to solve problems in healthcare. I'm also very interested in learning-to-optimize and automating scientific discovery.

Ongoing Research

Explicit Reasoning for Time Series Analysis

Training LLMs to perform explicit reasoning for time series analysis via reinforcement learning with verifiable rewards.

Paper ¶ Abstract

SimplexRL: Learning Optimal Pivot Selection for the Simplex Algorithm

Training a foundation model for pivot variable selection in the simplex algorithm using reinforcement learning.

Demo ¶ Abstract

Predictive Hospital Capacity Management

Building models to predict surge demand and proactively optimize patient flow at the Johns Hopkins Hospital.

Personalized DASH Diet Recommendations for Patients with Hypertension

Developing goal-oriented LLM-based recommender systems for helping patients with hypertension adhere to the DASH diet.

Demo

Completed Projects

Time-Series Augmented Multimodal Large Language Models

Developed a multimodal LLM framework for time series prediction and analysis. Published in ICML 2026.

Paper ICML Poster ¶ Abstract

MedTsLLM: Leveraging LLMs for Multi-Modal Medical Time Series Analysis

Built models for analyzing medical time series data that leverage LLMs for contextual understanding. Published in MLHC 2024.

Paper Code ¶ Abstract

Optimal Hospital Capacity Management During Demand Surges

Developed models to optimize practical capacity management decisions in hospital systems during surge periods. Published in Health Care Management Science.

Paper Code ¶ Abstract

Supervised Inverse Optimization

Extended inverse optimization to incorporate outcome data or scores.

¶ Abstract

screenshot of options in a dashboard

An Interactive Decision-Support Dashboard for Optimal Hospital Capacity Management

Built an interactive online dashboard to assist hospitals in making capacity management decisions during major demand surges.

Paper Demo Code ¶ Abstract

diagram of transfers between hospitals

Hospital Resource and Demand Redistribution for COVID-19

Developed models to make optimal patent and resource transfers between hospitals to reduce the burden of COVID-19 on healthcare systems.

Demo Paper Code ¶ Abstract

plot of forecast evaluation scores across locations and time

Evaluating Epidemic Forecasting

Introduced the weighted contextual interval score, a new metric for evaluating quantile forecasts in context.

Paper ¶ Abstract

diagram of matching graph nodes

Phantom Alignment Strength in Weakly Correlated Graphs

Studied alignment strength of seeded graph matching for weakly-correlated graphs.

Paper Code ¶ Abstract

image of a retina

Detecting Sickle-Cell Retinopathy

Used deep learning to detect sickle-cell retinopathy from ultra-widefield fundus photography.

Paper Code ¶ Abstract

diagram of segmented regions of a larynx

Automated Classification and Segmentation of Laryngeal Lesions

Instance segmentation for detecting two types of laryngeal lesions in endoscopic video.

Paper Code ¶ Abstract

image of a caataract surgery being performed

Cataract Surgery Phase Identification

Contributed to a machine learning system for recognizing the current phase in cataract surgery from video as a step towards automated skill assessment.

Paper ¶ Abstract

AlphaTSP

Built a heuristic solver for the traveling salesman problem using Monte Carlo tree search and deep reinforcement learning, inspired by AlphaZero.

Code Paper ¶ Abstract

Contact

[email protected]