Naveen Jindal School of Management · UT Dallas
Ph.D. Candidate in Information Systems, advised by Prof. Vijay Mookerjee. On the academic job market, 2026–27.
I study the design and operation of digital marketplaces — settings where a platform allocates scarce resources and sets fees while advertisers, sellers, and buyers respond strategically. My current work spans real-time bidding in digital advertising, the comparative design of closed platforms and open commerce networks, cost-effective routing in AI systems, and cyberinsurance. I build analytical and computational models using game theory, optimization, stochastic control, and machine learning.
Research
Digital advertising markets, platform and network design, and the economics of risk and cost in AI-enabled systems.
Digital billboards sell ad slots by real-time auction, but their audiences are hyper-local: the ads on a screen shape who bids for it next. I characterize the optimal policy for a screen owner who must split scarce inventory between auction revenue today and self-advertising that recruits tomorrow's bidders, and calibrate the model on transaction-level auction data.
The win-curve maps a bid to the probability of winning an impression, and nearly every bidding algorithm takes it as given. I derive it from primitives for first-price auctions with budget-constrained bidders, give a stable procedure to compute it, and show that budget-aware reserve prices recover revenue a budget-agnostic policy leaves behind.
CIST 2024 · INFORMS 2024 · Big XII+ MIS Research Symposium 2025 · POMS 2025
India's ONDC splits the functions of an e-commerce platform into independent buyer-side and seller-side apps. I show that decoupling alone leaves everyone worse off through double marginalization — and that the gains appear only when buyer-app entry stays open and sellers can run their own seller apps.
CIST 2025 · INFORMS 2025 · DSI 2025
Firms often absorb a breach rather than file a claim, fearing reputational damage. I ask whether an insurer that prices with this claiming behavior in mind can lower what firms pay while protecting its own revenue — making coverage accessible to firms currently priced out.
Answering a user query by retrieval is cheap; reasoning with a large model is accurate but costly. This project studies how a chatbot operator should route between them to balance service quality against the cost of serving.
Teaching
Before this
I came to Information Systems from the laboratory bench, which is why I like problems that need both a model and data.