Contact me at chohyu01 (at) cs.washington.edu.

Experience

• Senior Systems Software Engineer: I am part of the RAPIDS team, who builds RAPIDS, the cutting-edge software suite for accelerating machine learning on NVIDIA GPUs.
  NVIDIA. April 2020—Present.
• Applied Scientist: I was part of initial efforts to launch Amazon SageMaker Neo in November 2018.
  Amazon Web Services. January 2018—February 2020.
• Applied Scientist Intern
  Amazon Web Services. June—December 2017.

Open-source Involvement

• Lead Maintainer of XGBoost
  XGBoost is an efficient, scalable library for gradient boosting. Over the years, I made substantial contribution to the C++ / Python portion of XGBoost. I also built and have maintained a continuous integration / continuous delivery (CI/CD) server. The server is hosted on AWS and runs a variety of tests for all incoming code contributions, ensuring that XGBoost works well with multiple operating systems, data infrastructures (Dask, Apache Spark) and NVIDIA GPUs. In addition, I made multiple renovations for the build and packaging system used in XGBoost, ensuring that XGBoost works well with other packages in Python and C++ ecosystem.
• Owner and Maintainer of Treelite
  Treelite is a model compiler for easy and efficient deployment of decision tree ensembles. The Treelite project is used within Amazon SageMaker Neo and RAPIDS Machine Learning Library.
• My GitHub account

Education

• M.S. in Computer Science
  Paul G. Allen School of Computer Science and Engineering, University of Washington. September 2015—March 2018.
• B.S. in Computer Science and Mathematics
  Trinity College. 2012—2015.

Peer-reviewed publications

• Survival regression with accelerated failure time model in XGBoost
  with Avinash Barnwal and Toby D. Hocking
  Article published in Journal of Computational and Graphical Statistics (2022)
• Block-distributed Gradient Boosted Trees
  with Theodore Vasiloudis and Henrik Boström
  Paper presented at ACM SIGIR (2019), best short paper award
• Treelite: toolbox for decision tree deployment
  with Mu Li
  Poster presented at SysML 2018 (2018)
• An accelerated procedure for hypergraph coarsening on the GPU
  with Lin Cheng and Peter Yoon
  Paper presented at IEEE HPEC (2015)
• A memory-efficient algorithm for large-scale symmetric tridiagonal eigenvalue problem on multi-GPU systems
  with Peter Yoon
  Paper presented at WorldComp 2014: PDPTA (2014)
  Poster presented at GPU Technology Conference (2014)
• GPU accelerated vessel segmentation using Laplacian eigenmaps
  with Lin Cheng, Peter Yoon and Jiajia Zhao
  Paper presented at IASTED PDCN (2014)
  Poster presented at GPU Technology Conference (2013)
• An efficient out-of-core implementation of block Cholesky decomposition on a multi-GPU system
  with Lin Cheng, Peter Yoon and Jiajia Zhao
  Paper presented at IASTED PDCS (2012), best paper award
  Poster presented at IEEE EMBS (2012)

Talks

• The State of XGBoost: history and community overview
  Los Angeles Data Science Meetup, November 10, 2020
• Taming Complexity in Software: real-life stories and lessons
  Trinity College Senior Seminar Guest Talk, October 21, 2020

Awards and Honors

• Best Short Paper Award, ACM SIGIR 2019, Paris, France, July 25, 2019.
• Winner of Outstanding Undergraduate Researcher Award, Computing Research Association, 2015.
• Recipient of the Goldwater Scholarship, The United States Congress, 2014.
• Best Paper Award, The 24th IASTED International Conference on Parallel and Distributed Computing and System, Las Vegas, NV, November 13, 2012.