I am a reliable, versatile, creative, entrepreneurial young independent scientist and content creator specialized in statistical learning, distributed optimization, high-performance computing, and package development.

I believe that machine learning should not be a “rich” man’s sport, and I am working to apply machine learning to any aspect of life and to make it accessible to everyone.

Besides my expertise on mathematics and computer science, I have also abundant knowledge on economics, finance, and sociology.

In addition to doing research and programming, I also like playing Yugioh Duel Links casually. With my original Hazy Flame deck, I achieved the following:

  • 1st KC Cup of 2018: 47653rd globally;
  • Season 26: King of Games (~1% of the population);
  • World Championship 2018 Qualifiers: 2529th Europe.

Current Projects

  • Next-generation distributed optimization
  • Fundamental machine learning questions

Past Projects

  • Distributed Frank–Wolfe
  • Private and Decentralized recommender systems
  • Clustering of flat steel surface defects
  • Calibration of Levy process

Research Interests

  • Nonparametric statistics
  • First-order convex optimization
  • Distributed machine learning
  • Recommender systems
  • Differential privacy & fairness
  • Adaptive data analysis


Reference Card


  • Learning Low-rank Matrices Distributedly without Factorization
  • Pokémon GO, Metagaming, and Data Science


A Distributed Frank–Wolfe Framework for Trace Norm Minimization via the Bulk Synchronous Parallel Model



  • Neurons – a concise Python package implementing a Neural Network without hidden layers.
  • Breakout – a simple game on Windows, written in C++.
  • From 2048 to ABCD – a 2048-like game, open-source, written in C++.
  • Multithreading of C++ 11– a document testing C++11 multithreading technology.
  • Park – a Python package of PySpark offline testing environment.
  • Vocabulary Memory – an Excel file helping you memorize vocabulary. You can also customize it to memorize other things. Should enable the macro.
  • Fate_GrandOrder – an IPython notebook to help you visualize the stats growth curves in Fate / Grand Order.