Saad Mohammad Abrar

Introduction

Hi, I am Saad! I am a PhD candidate in Computer Science at the University of Maryland, College Park, working under the supervision of Dr. Vanessa Frias-Martinez in the Urban Computing Lab.

My research is in general is situated in computational social science, drawing methodologies from machine learning, statistics, and fairness methods. I aim to improve high-stakes decision-making by leveraging behavioral data to build accurate and trustworthy predictive systems.

My dissertation, Data-Driven Decision-Making for Societal Systems: Auditing Mobility, Contextual Data and Models for High-Stakes Applications, develops frameworks to audit mobility and contextual data—and the predictive models built on them—along the dimensions of utility, fairness, and validity before they are trusted in high-stakes settings such as public health and economic policy. Through my research, I aim to ensure that data-driven urban computing applications serve all communities equitably while maintaining high predictive performance. My research has led to 16 peer-reviewed publications (10 journal articles, 2 conference papers, 4 workshop papers) in venues such as PLOS One, EMNLP, International Journal of Social Research Methodology, ACM SIGSPATIAL, Computers Environment and Urban Systems, ACM JCSS, EPJ Data Science, Big Data & Society, and Behavior Research Methods.

I completed my M.S. in Computer Science from UMD (2023) and my BSc in Computer Science and Engineering from Bangladesh University of Engineering and Technology (BUET) in 2017.

Dissertation

Research Statement

Teaching Statement

Academic CV

Research Focus

Research Framework

Model: I design accurate predictors that leverage contextual and behavioral signals to improve forecasting in high-stakes settings. Through systematic evaluation, I determine when and how mobility and social media data enhance predictions across public health, crime, and related domains. [EPJ Data Science ‘22, JCSS ‘23, JMIR ‘23]

Mitigate: I develop comprehensive audit frameworks that uncover bias and failure modes in both data and models. By separating data-driven from model-driven failures, I create targeted fairness strategies that address disparities across demographics and geographies in predictive systems. [CEUS ‘23, JCSS ‘23, PLoS One ‘25, EMNLP Findings ‘25]

Measure: I build ethically-aware data collection toolkits that empower communities to gather and analyze their own mobility and experiential data. Through participatory design that prioritizes consent and transparency, I enable equitable decision-making in public transit planning. [COMPASS ‘23, IJSRM ‘25]

This research aims to enable efficient and equitable data-driven policy and planning, with direct applications in:

Public Health Response
Urban Planning
Economic Recovery
Transportation Equity