Xiangyu Zhang 

Email: xz556@cornell.edu 607-229-0792 
Education 
Cornell University, Ithaca, NY 
Ph.D. in Operations Research and Information Engineering. Minor: Computer Science, Statistics. 
Aug 2017 -Present 

• Coursework: Supervised Learning, Unsupervised Learning, Machine Learning Theory, Linear Model, Statistical 
Inference, Algorithm, Stochastic Process, Stochastic Calculus, Operating System. 

Tsinghua University, Beijing Aug 2013 -Jul 2017 

B.S. in Mathematics, GPA: 91 / 100. 

Awards 

Silver Medal in CMO (Chinese Mathematical Olympiad) 2012 
Silver Medal in CMO (Chinese Mathematical Olympiad) 2013 

Skills 

• Programming Language: Python, Bash, C. 
• Machine Learning: Clustering, SVM, HMM, Decision Tree, Neural Network. 
• Familiar with numpy, pandas, sklearn, statsmodels and pytorch. 


Work Experiences 

Algo Developer, Hudson River Trading May 2021 -Aug 2021 

• Designed trading signals for earning announcement events. One signal is lively traded and generates ∼250K proÿt. 

• Estimated the distribution of hidden liquidity within spread for US equities. 
Quant Researcher, Susquehanna International Group May 2020 -Aug 2020 

• Evaluated various Barra factor models in term of portfolio volatility prediction. 
• Optimized portfolio under Markowitz framework. Boosted the optimization process 100x by exploiting structures 
of factor models. 


Research Scientist, IBM May 2018 -Aug 2018 

• Developed machine learning algorithms for predicting the probability of winning IT service contracts, achieving 
accuracy higher than 90%. 
• Priced highly valued IT service contracts using reinforcement learning. 



Research Experiences 

Resource allocation in Weakly Coupled Markov Decision Processes (WCMDP). 

• Developed a near-optimal policy for WCMDP with strong theoretical guarantees. 

• Showed the ÿrst known regret bound that does not grow with the number of arms. 
Analysis of multi-class queueing network under static priority policies. 

• Developed a framework to uniformly bound higher-priority classes’ queue length regardless of tra…c intensities. 
• Showed this uniform bound for Last/First-Bu•er-First-Serve policies. 


Working Papers 

Xiangyu Zhang & Peter Frazier: Restless Bandits with Many Arms: Beating the Central Limit Theorem. Submitted to 
Operation Research. https://arxiv.org/pdf/2107.11911.pdf 

Xiangyu Zhang & Peter Frazier: Near-Optimality for Inÿnite-Horizon Restless Bandits with Many Arms. Ready to submit. 

Xiangyu Zhang & Peter Frazier: Near-optimality for Finite-Horizon Weakly Coupled Markov Decision Processes with 
Many Arms. Ready to submit. 

Yunxiang Zhang, Xiangyu Zhang & Peter Frazier: Two-Step Lookahead Bayesian Optimization with Inequality Constraints. 
Submitted to AISTATS. 

Xiangyu Zhang, Aly Megahed & Peter Frazier: Dynamic Pricing with Long-Term Relationships. Submitted to Management 
Science. 

Chang Cao, Jim Dai, Masakiyo Miyazawa & Xiangyu Zhang: State Space Collapse for Multi-class Queueing Networks under 
SBP Service Policies. 

Selected Presentations 
INFORMS Revenue Management and Pricing Section Conference Toronto, June 2018 
INFORMS Annual Meeting Phoenix, Oct 2018 
INFORMS Annual Meeting Seattle, Oct 2019 
University of Washington (invited by Professor Archis Ghate) Seattle, Oct 2019 
Georgia Institute of Technology (invited by Professor Enlu Zhou) Atlanta, Dec 2019 

2