Tao Lin
Tao Lin

PhD in Computer Science

Harvard University

About Me

Contact: tlin@g.harvard.edu or lintao@cuhk.edu.cn

I am a postdoctoral researcher at Microsoft Research (New England), in the Economics and Computation group, hosted by Alex Slivkins.

I obtained my PhD in Computer Science from Harvard University in 2025 (advised by Yiling Chen) and BSc from Peking University in 2020 (advised by Xiaotie Deng). My research spans economics, machine learning, and theoretical computer science. I am especially interested in mechanism design and information design for learning-based decision-makers, with applications to, e.g., advertising auctions and recommender systems. From 2023 to 2024, I interned at ByteDance and Google. I received the Siebel Scholar Award in 2025.

I will be an assistant professor in the School of Data Science at the Chinese University of Hong Kong, Shenzhen, starting in 2026. I am looking for PhD students starting in Fall 2026; please see here for details.

Interests
  • Algorithmic Game Theory
  • Mechanism Design
  • Machine Learning
  • Theoretical Computer Science
Education

  • PhD in Computer Science

    Harvard University

  • BSc in EECS

    Peking University

🗞️ News
📚 My Research

My research direction is learning-based incentive design, an interdisciplinary topic in economics, machine learning, and theoretical computer science. I study mechanism design and information design problems with learning-based decision-makers. Example directions include:

  • Learning agents: I investigate how the learning behavior of boundedly rational agents (modeled by, e.g., reinforcement learning) affects the outcome of games, compared to the classical outcome predicted by rational-agent-based economic theory.
  • Learning principals: I also study how the principals (designers of mechanisms and information structures) can achieve optimal design goals by learning unknown parameters about agents and environments from repeated interactions. Such learning problems involving dynamic and strategic data sources, departing from traditional machine learning paradigms, requiring new methodologies that I aim to develop.

My research is often motivated by the interplay between economic incentives and machine learning algorithms in real-world AI systems, such as advertising auctions and recommender systems.

📑 Featured Projects
Information Design With Large Language Models
Information Design

Information Design With Large Language Models

Information design is typically studied through the lens of Bayesian signaling, where signals shape beliefs based on their correlation with the true state of the world. However, Behavioral Economics and Psychology emphasize that human decision-making is more complex and can depend on how information is framed. This paper formalizes a language-based notion of framing and bridges this to the popular Bayesian-persuasion model. We model framing as a possibly non-Bayesian, linguistic way to influence a receiver’s belief, while a signaling (or recommendation) scheme can further refine this belief in the classic Bayesian way. A key challenge in systematically optimizing in this framework is the vast space of possible framings and the difficulty of predicting their effects on receivers. Based on growing evidence that Large Language Models (LLMs) can effectively serve as proxies for human behavior, we formulate a theoretical model based on access to a framing-to-belief oracle. This model then enables us to precisely characterize when solely optimizing framing or jointly optimizing framing and signaling is tractable. We substantiate our theoretical analysis with an empirical algorithm that leverages LLMs to (1) approximate the framing-to-belief oracle, and (2) optimize over language space using a hill-climbing method. We apply this to two marketing-inspired case studies and validate the effectiveness through analytical and human evaluation.

Generalized Principal-Agent Problem with a Learning Agent
Information Design

Generalized Principal-Agent Problem with a Learning Agent

Classic principal-agent problems such as Stackelberg games, contract design, and Bayesian persuasion, often assume that the agent is able to best respond to the principal’s committed strategy. We study repeated generalized principal-agent problems under the assumption that the principal does not have commitment power and the agent uses algorithms to learn to respond to the principal. We reduce this problem to a one-shot generalized principal-agent problem where the agent approximately best responds. Using this reduction, we show that: (1) If the agent uses contextual no-regret learning algorithms with regret $\mathrm{Reg}(T)$, then the principal can guarantee utility at least $U^* - \Theta\big(\sqrt{\tfrac{\mathrm{Reg}(T)}{T}}\big)$, where $U^*$ is the principal’s optimal utility in the classic model with a best-responding agent. (2) If the agent uses contextual no-swap-regret learning algorithms with swap-regret $\mathrm{SReg}(T)$, then the principal cannot obtain utility more than $U^* + O(\frac{\mathrm{SReg(T)}}{T})$. But (3) if the agent uses mean-based learning algorithms (which can be no-regret but not no-swap-regret), then the principal can sometimes do significantly better than $U^*$. These results not only refine previous results in Stackelberg games and contract design, but also lead to new results for Bayesian persuasion with a learning agent and all generalized principal-agent problems where the agent does not have private information.

User-Creator Feature Polarization in Recommender Systems with Dual Influence
Recommender System

User-Creator Feature Polarization in Recommender Systems with Dual Influence

Recommender systems serve the dual purpose of presenting relevant content to users and helping content creators reach their target audience. The dual nature of these systems naturally influences both users and creators: users’ preferences are affected by the items they are recommended, while creators may be incentivized to alter their content to attract more users. We define a model, called user-creator feature dynamics, to capture the dual influence of recommender systems. We prove that a recommender system with dual influence is guaranteed to polarize, causing diversity loss in the system. We then investigate, both theoretically and empirically, approaches for mitigating polarization and promoting diversity in recommender systems. Unexpectedly, we find that common diversity-promoting approaches do not work in the presence of dual influence, while relevancy-optimizing methods like top-$k$ truncation can prevent polarization and improve diversity of the system.

Nash Convergence of Mean-Based Learning Algorithms in First Price Auctions
Multi-Agent Learning in Auctions

Nash Convergence of Mean-Based Learning Algorithms in First Price Auctions

The convergence properties of learning dynamics in repeated auctions is a timely and important question, with numerous applications in, e.g., online advertising markets. This work focuses on repeated first-price auctions where bidders with fixed values learn to bid using mean-based algorithms — a large class of online learning algorithms that include popular no-regret algorithms such as Multiplicative Weights Update and Follow the Perturbed Leader. We completely characterize the learning dynamics of mean-based algorithms, under two notions of convergence: (1) time-average: the fraction of rounds where bidders play a Nash equilibrium converges to 1; (2) last-iterate: the mixed strategy profile of bidders converges to a Nash equilibrium. Specifically, the results depend on the number of bidders with the highest value:

📃 Working Papers
On the Coordination of Value-Maximizing Bidders
Yanru Guan, Jiahao Zhang, Zhe Feng, Tao Lin
Working paper
PDF Cite
Information Design With Large Language Models
(Alphabet) Paul Duetting, Safwan Hossain, Tao Lin, Renato Paes Leme, Sai Srivatsa Ravindranath, Haifeng Xu, Song Zuo
Working paper
PDF Cite Slides
Learning to Play Multi-Follower Bayesian Stackelberg Games
Gerson Personnat, Tao Lin, Safwan Hossain, David C. Parkes
Working paper
PDF Cite
WOMAC: A Mechanism For Prediction Competitions
Siddarth Srinivasan, Tao Lin, Connacher Murphy, Anish Thilagar, Yiling Chen, Ezra Karger
Working paper
PDF Cite
Explainable Information Design
(Alphabet) Yiling Chen, Tao Lin, Wei Tang, Jamie Tucker-Foltz
Working paper
PDF Cite
Learning to Coordinate Bidders in Non-Truthful Auctions
(Alphabet) Hu Fu, Tao Lin
Working paper
PDF Cite
Learning a Game by Paying the Agents
Brian Hu Zhang*, Tao Lin*, Yiling Chen, Tuomas Sandholm
Working paper
PDF Cite
📃 Publications
See here.
📝 Notes
How Does Independence Help Generalization? Sample Complexity of ERM on Product Distributions
Tao Lin
Cite URL
On Clearing Prices in Matching Markets: A Simple Characterization without Duality
(Alphabet) Xiaoming Li, Tao Lin
Cite URL
🎙️ Recent & Upcoming Talks
See all

Experience

  1. Assistant Professor

    The Chinese University of Hong Kong, Shenzhen

  2. Postdoctoral Researcher

    Microsoft
    Hosted by Alex Slivkins

  3. Student Researcher

    Google
    Hosted by Christopher Liaw

  4. Research Intern

    ByteDance
    Hosted by Yang Liu

Education

  1. PhD in Computer Science

    Harvard University
    Dissertation: Incentive Design in the Machine Learning Age

  2. BSc in EECS

    Peking University
For Prospective Students

I am looking for PhD and MPhil students starting in Fall 2026. If you have:

  • strong interest in pursuing interdisciplinary research across Computer Science, Data Science, and Economics;
  • good mathematical background and coding skills;
  • (bonus) ample experience in prompting GenAI for research and judging whether their responses are correct;

then please consider applying to the MPhil-PhD Program in Data Science or Computer Science of CUHK-Shenzhen, SDS, and mention my name in your application. Feel free to contact me at lintao@cuhk.edu.cn