About

Krishnamurthy Iyer is an Associate Professor and the Director of Masters Programs at the Department of Industrial and Systems Engineering at the University of Minnesota. He received his PhD from the Department of Management Science and Engineering at Stanford University in 2012, and his B.Tech and M.Tech in Mechanical Engineering from the Indian Institute of Technology, Bombay in 2006. He spent a year as a Postdoctoral Researcher at the University of Pennsylvania's Computer and Information Science Department. His research interests include game theory and stochastic modeling with applications in the design and analysis of online markets and service systems. His current research focuses on studying optimal mechanisms for information sharing in service systems and platform markets.

Research topics

• Computer Science
• Economics
• Business
• Macroeconomics
• Engineering
• Mathematical economics
• Microeconomics
• Operations management

Selected publications

• How to Sell a Service with Uncertain Outcomes

  2025-07-02

  articleOpen access1st authorCorresponding

  Motivated by the recent popularity of machine learning training services, we introduce a contract design problem in which a provider sells a service that results in an outcome of uncertain quality for the buyer. The seller has a set of actions that lead to different distributions over outcomes. We focus on a setting in which the seller has the ability to commit to an action and the buyer is free to accept or reject the outcome after seeing its realized quality. Our model is related to Mussa and Rosen's classic paper on selling products of differing qualities and monopolist lottery pricing, as well as recent work on selling hidden actions.

  PublisherOA PDFDOI

• How to Sell a Service with Uncertain Outcomes

  ArXiv.org · 2025-02-18

  preprintOpen access1st authorCorresponding

  Motivated by the recent popularity of machine learning training services, we introduce a contract design problem in which a provider sells a service that results in an outcome of uncertain quality for the buyer. The seller has a set of actions that lead to different distributions over outcomes. We focus on a setting in which the seller has the ability to commit to an action and the buyer is free to accept or reject the outcome after seeing its realized quality. We propose a two-stage payment scheme where the seller designs a menu of contracts, each of which specifies an action, an upfront price and a vector of outcome-dependent usage prices. Upon selecting a contract, the buyer pays the upfront price, and after observing the realized outcome, the buyer either accepts and pays the corresponding usage price, or rejects and is exempt from further payment. We show that this two-stage payment structure is necessary to maximize profit: only upfront prices or only usage prices is insufficient. We then study the computational complexity of computing a profit-maximizing menu in our model. While computing the exact maximum seller profit is NP-hard even for two buyer types, we derive a fully-polynomial time approximation scheme (FPTAS) for the maximum profit for a constant number of buyer types. Finally, we prove that in the single-parameter setting in which buyers' valuations are parametrized by a single real number that seller revenue can be maximized using a menu consisting of a single contract.

  PublisherOA PDFDOI

• Learning to Persuade on the Fly: Robustness Against Ignorance

  Operations Research · 2024-06-18 · 3 citations

  article

  How Can Platforms Learn to Make Persuasive Recommendations? Many online platforms make recommendations to users on content from creators or products from sellers. The motivation underlying such recommendations is to persuade users into taking actions that also serve system-wide goals. To do this effectively, a platform needs to know the underlying distribution of payoff-relevant variables (such as content or product quality). However, this distributional information is often lacking—for example, when new content creators or sellers join a platform. In “Learning to Persuade on the Fly: Robustness Against Ignorance,” Zu, Iyer, and Xu study how a platform can make persuasive recommendations over time in the absence of distributional knowledge using a learning-based approach. They first propose and motivate a robust-persuasiveness criterion for settings with incomplete information. They then design an efficient recommendation algorithm that satisfies this criterion and achieves low regret compared with the benchmark of complete distributional knowledge. Overall, by relaxing the strong assumption of complete distributional knowledge, this research extends the applicability of information design to more practical settings.

  PublisherDOI

• Markov Persuasion Processes with Endogenous Agent Beliefs

  arXiv (Cornell University) · 2023-07-06

  preprintOpen access1st authorCorresponding

  We consider a dynamic Bayesian persuasion setting where a single long-lived sender persuades a stream of ``short-lived'' agents (receivers) by sharing information about a payoff-relevant state. The state transitions are Markovian and the sender seeks to maximize the long-run average reward by committing to a (possibly history-dependent) signaling mechanism. While most previous studies of Markov persuasion consider exogenous agent beliefs that are independent of the chain, we study a more natural variant with endogenous agent beliefs that depend on the chain's realized history. A key challenge to analyze such settings is to model the agents' partial knowledge about the history information. We analyze a Markov persuasion process (MPP) under various information models that differ in the amount of information the receivers have about the history of the process. Specifically, we formulate a general partial-information model where each receiver observes the history with an $\ell$ period lag. Our technical contribution start with analyzing two benchmark models, i.e., the full-history information model and the no-history information model. We establish an ordering of the sender's payoff as a function of the informativeness of agent's information model (with no-history as the least informative), and develop efficient algorithms to compute optimal solutions for these two benchmarks. For general $\ell$, we present the technical challenges in finding an optimal signaling mechanism, where even determining the right dependency on the history becomes difficult. To bypass the difficulties, we use a robustness framework to design a "simple" \emph{history-independent} signaling mechanism that approximately achieves optimal payoff when $\ell$ is reasonably large.

  PublisherOA PDFDOI

• Persuading Risk-Conscious Agents: A Geometric Approach

  Operations Research · 2023-03-28 · 8 citations

  article

  A Convex Programming Framework for Information Design Under Realistic Human Behavior Platform markets and services typically have additional relevant information in comparison with their users. Information design studies how the sharing of this information can be leveraged by the platform to influence user behavior and obtain desirable outcomes. Previous research has studied information design assuming that the users act to maximize their expected utility, but this assumption does not always hold in reality. Instead, people often exhibit biases and deviations from expected utility maximization. In “Persuading Risk-Conscious Agents: A Geometric Approach,” Anunrojwong, Iyer, and Lingenbrink study information design with “risk-conscious” agents whose utility functions may depend nonlinearly on their beliefs. They provide a convex programming approach for solving for the optimal persuasion mechanism and establish their structural properties in different settings. They illustrate their approach in an application involving the sharing of waiting-time information in a queueing system. Overall, this work contributes to the study of information design under realistic models of human behavior.

  PublisherDOI

• Information Design for Congested Social Services: Optimal Need-Based Persuasion

  Management Science · 2022 · 36 citations

  • Computer Science
  • Business
  • Computer Science

  We study the effectiveness of information design in reducing congestion in social services catering to users with varied levels of need. In the absence of price discrimination and centralized admission, the provider relies on sharing information about wait times to improve welfare. We consider a stylized model with heterogeneous users who differ in their private outside options: low-need users have an acceptable outside option to the social service, whereas high-need users have no viable outside option. Upon arrival, a user decides to wait for the service by joining an unobservable first-come-first-serve queue, or leave and seek her outside option. To reduce congestion and improve social outcomes, the service provider seeks to persuade more low-need users to avail their outside option, and thus better serve high-need users. We characterize the Pareto-efficient signaling mechanisms and compare their welfare outcomes against several benchmarks. We show that if either type is the overwhelming majority of the population, then information design does not provide improvement over sharing full information or no information. On the other hand, when the population is sufficiently heterogeneous, information design not only Pareto-dominates full-information and no-information mechanisms, in some regimes it also achieves the same welfare as the “first-best,” that is, the Pareto-efficient centralized admission policy with knowledge of users’ types. This paper was accepted by Gabriel Weintraub, revenue management and market analytics. Funding: This work was supported by the National Science Foundation, Division of Civil, Mechanical and Manufacturing Innovation [Grants CMMI-2002155 and CMMI-2002156]. Supplemental Material: The data and e-companion are available at https://doi.org/10.1287/mnsc.2022.4548 .

  PublisherDOI

• Persuading Risk-Conscious Agents: A Geometric Approach

  arXiv (Cornell University) · 2022-08-07

  preprintOpen access

  We consider a persuasion problem between a sender and a receiver whose utility may be nonlinear in her belief; we call such receivers risk-conscious. Such utility models arise when the receiver exhibits systematic biases away from expected-utility-maximization, such as uncertainty aversion (e.g., from sensitivity to the variance of the waiting time for a service). Due to this nonlinearity, the standard approach to finding the optimal persuasion mechanism using revelation principle fails. To overcome this difficulty, we use the underlying geometry of the problem to develop a convex optimization framework to find the optimal persuasion mechanism. We define the notion of full persuasion and use our framework to characterize conditions under which full persuasion can be achieved. We use our approach to study binary persuasion, where the receiver has two actions and the sender strictly prefers one of them at every state. Under a convexity assumption, we show that the binary persuasion problem reduces to a linear program, and establish a canonical set of signals where each signal either reveals the state or induces in the receiver uncertainty between two states. Finally, we discuss the broader applicability of our methods to more general contexts, and illustrate our methodology by studying information sharing of waiting times in service systems.

  PublisherOA PDFDOI

• Learning to Persuade on the Fly: Robustness Against Ignorance

  arXiv (Cornell University) · 2021-02-19

  preprintOpen access

  Motivated by information sharing in online platforms, we study repeated persuasion between a sender and a stream of receivers where at each time, the sender observes a payoff-relevant state drawn independently and identically from an unknown distribution, and shares state information with the receivers who each choose an action. The sender seeks to persuade the receivers into taking actions aligned with the sender's preference by selectively sharing state information. However, in contrast to the standard models, neither the sender nor the receivers know the distribution, and the sender has to persuade while learning the distribution on the fly. We study the sender's learning problem of making persuasive action recommendations to achieve low regret against the optimal persuasion mechanism with the knowledge of the distribution. To do this, we first propose and motivate a persuasiveness criterion for the unknown distribution setting that centers robustness as a requirement in the face of uncertainty. Our main result is an algorithm that, with high probability, is robustly-persuasive and achieves $O(\sqrt{T\log T})$ regret, where $T$ is the horizon length. Intuitively, at each time our algorithm maintains a set of candidate distributions, and chooses a signaling mechanism that is simultaneously persuasive for all of them. Core to our proof is a tight analysis about the cost of robust persuasion, which may be of independent interest. We further prove that this regret order is optimal (up to logarithmic terms) by showing that no algorithm can achieve regret better than $Ω(\sqrt{T})$.

  PublisherOA PDFDOI

• The Remarkable Robustness of the Repeated Fisher Market

  2021-07-18 · 3 citations

  articleOpen accessSenior author

  In many settings, resources are allocated among agents repeatedly over time without the use of monetary transfers: consider, for example, allocating server-time to company employees, rooms to students, or food among food banks. Here, the central challenge is to allocate resources efficiently despite the absence of payments. In this work we study a simple online variant of the standard Fisher market, where we endow all agents with a budget of artificial credits, and then repeatedly run simultaneous first-price auctions for each item in each period. Owing to their simplicity, such mechanisms have been gaining in popularity, with several recent successful implementations, most notably, by Feeding America for US food banks. Our goal in this paper is to understand the incentive and efficiency properties of these mechanisms.

  PublisherOA PDFDOI

• Information Design for Congested Social Services: Optimal Need-Based Persuasion

  SSRN Electronic Journal · 2021-01-01

  preprintOpen access
  PublisherDOI

Recent grants

• Repeated Auctions in Incomplete Information Settings with Learning Bidders

  NSF · $202k · 2017–2019

• Pricing Services and Products in the Presence of Parallel Delivery Mechanisms and Strategic Consumer Behavior

  NSF · $31k · 2019–2020

• Pricing Services and Products in the Presence of Parallel Delivery Mechanisms and Strategic Consumer Behavior

  NSF · $300k · 2015–2020

• Repeated Auctions in Incomplete Information Settings with Learning Bidders

  NSF · $9k · 2019–2021

Frequent coauthors

• Ramesh Johari

  13 shared

• Siddhartha Banerjee

  Cornell University

  12 shared

• Artur Gorokh

  12 shared

• David Lingenbrink

  Cornell University

  11 shared

• Jerry Anunrojwong

  Columbia University

  10 shared

• Ciamac C. Moallemi

  7 shared

• Vahideh Manshadi

  Yale University

  6 shared

• M. Sundararajan

  Mizoram University

  6 shared

Education

• PhD, Management Science and Engineering

  Stanford University

  2012

Awards & honors

• First Place, INFORMS Junior Faculty Interest Group (JFIG) Pa…
• Ralph S. Watts ’72 Excellence in Teaching Award, College of…
• Dantzig-Lieberman Fellowship (Stanford University) 2011
• Stanford Graduate Fellowship (Office of Technology Licensing…
• Institute Gold Medal (IIT Bombay) 2006