My name is Peng Shi and I am a post-doctoral researcher at Microsoft Research New England. I will be starting a tenure-track position at USC Marshall in the Data Science and Operations group in June 2017.
I am interested in developing mathematical models and techniques that can significantly benefit society. My current focus is prediction and optimization in matching markets, which include systems that match students to schools, applicants to subsidized housing, workers to jobs, and organ donors to recipients. Characteristics of these systems include heterogeneous supply and demand, potential strategic behavior of agents, and inability of obtaining the desired allocation by only setting prices. My PhD thesis, "Prediction and optimization in school choice", was motivated by school choice in Boston, for which I proposed an assignment plans that was adopted in March 2013 (for news coverage, see Boston Globe 2012/10/27, NY Times 2013/3/13, and NY Times 2013/3/15). I am currently working on developing theories and methodologies to assist future school choice reforms, and on applying optimization to other matching markets.
Assortment Planning in School Choice. (Preliminary Draft).
Optimal Allocation without Money: an Engineering Approach (with Itai Ashlagi). Management Science, Forthcoming, 2015.
- Won the 2014 MIT Operations Research Center Best Student Paper Competition.
- Won the 2013 INFORMS Public Sector OR Best Paper Competition.
- An earlier version appeared in EC' 14.
Guiding School-Choice Reform through Novel Applications of Operations Research. Interfaces, 45(2), 2015.
- Won the 2013 INFORMS Doing Good with Good OR Best Paper Competition.
Improving Community Cohesion in School Choice via Correlated-Lottery Implementation (with Itai Ashlagi). Operations Research , 62(6), 2014.
Approximation algorithms for restless bandit problems (with Kamesh Munagala and Sudipto Guha). Journal of the ACM (JACM) , 58(1), 2010.
Refereed Conference Proceedings
Prediction Mechanisms that Do Not Incentivize Undesirable Actions (with Vincent Conitzer and Mingyu Guo). Appeared in WINE'09.
Approximation algorithms for restless bandit problems (with Kamesh Munagala and Sudipto Guha). Appeared in SODA'09.
MIT Sloan School of Management:
- Fall, 2013: Teaching assistant for 15.060 (Data, Models, and Decisions).
- Fall, 2009: Co-Instructor for Math149S (Problem Solving Seminar). See here for course materials I created.