At the time I started working on my dissertation at Illinois, the existing literature on aviation security modeling was divided into two streams: work on security devices and optimal system design, and work on optimal passenger screening. My key contribution to this field lies in designing models that allow for treating both of these decision-making problems simultaneously. While working on this application, I also made methodological advances in sequential decision-making under uncertainty and local search algorithm design.

​My expertise in designing and applying optimization algorithms then led to methodological developments in causal inference with observational data. The conventional observational causal inference approaches employed over the past forty years have exploited matching of treated and untreated (control) units, with the quality of matching evaluated using covariate balance measures. The two approaches I have worked on, Balance Optimization Subset Selection and Mutual Information-based Matching, directly optimize balance to achieve accurate and fast inference.

​The research in social network analysis addresses strives for prescriptive rather descriptive advances. My methods for understanding human behavior based on static and longitudinal data combine models of communication with models of information exchange. The projects on influence maximization, opinion formation and strategic distribution of incentives and social capital are done with the help of my PhD students and in collaboration with colleagues across multiple disciplines.

​My work on program/policy evaluation highlights the importance of "reach" dimension in the studies with connected human subjects. I argue that reach can be quantified and show that it can be manipulated via incentives and recommendations, informing targeted interventions in online pro-health forums and for fostering environmentally conscious behavior.

​Finally, my most recent interests are in the areas of educational data mining and collaborative learning. I am working to develop environments and practices that would make students more motivated and excited about learning, by letting them work in groups, challenging and helping each other.