Hi there, I'm James Xiang

James Xiang

I enjoy all creative pursuits in life. I specialize in using maths and technology to solve challenging problems. Currently I work for the quantitative hedge fund Citadel, focusing on creating automated trading strategies.

I live in Chicago, United States.

I have always been very passionate about both math and technology: I enjoyed playing with radios and PCBs as a child; competed in math Olympiad in high school; studied Electrical Engineering; and enjoyed the mathematical principle and rigor in my PhD research. I valued both the power of math in providing the guiding principles through abstractions of reality, and the engineering mindset of "making things work" through technology innovations. Both requires creative problem solving, and with a combination of the two, we can create powerful solutions to many challenging pratical problems.

Brief Academic/Professional Timeline:


Middle school

Partipated in the IMO in 2003 and won a gold medal.



B.E., Electrical Engineering

Tsinghua University

Ranked 1/164 in my class.


Graduate School

Ph.D., Electrical Engineering

M.A., Electrical Engineering

Princeton University

Awarded Honorific Fellowship


Working at Citadel

2012 ~ 2017: QR
2017 ~ 2019: Senior QR
2019 ~ now : QR Lead

My PhD Research

In my PhD research I focused on machine learning, more specifically, on sparse representations and dictionary learning.

I was interested in how to learn sparse representations efficiently for large scale, high dimensional data sets. I have engineered a framework to learn hierachical, tree-structured dictionaries. I also made important mathematical contributions on solving the lasso problem, a key component in dictionary learning. My contribution is a set of "screening tests", which are mathematically-proven sufficient conditions for regressors to receive 0-weights in the optimal solution of the lasso problem. Applying these screening tests allow us to remove regressors from the lasso problem without affecting the solution.

For more details on these research, please see the following two representative papers: