Stephen Tu
Research Scientist at Google
New York, New York, United States
Summary
🤩
Rockstar🎓
Top SchoolStephen Tu is a research scientist at Google with 16 years of experience bridging academic research and production-quality C++ systems, holding advanced degrees from MIT and a PhD from UC Berkeley. His work focuses on numerical optimization and machine learning infrastructure, including notable open-source contributions to mlpack and ensmallen where he implemented and iterated on sparse, low-rank semi-definite programming solvers. He has a strong research pedigree from MIT, Berkeley, and CSAIL and a history of turning theoretical algorithms into tested, header-only C++ libraries used by practitioners. Comfortable across research and engineering, he blends deep algorithmic insight with pragmatic back-end development to ship robust optimization tooling.
16 years of coding experience
10 years of employment as a software developer
BA and BS, Computer Science and Mechanical Engineering, BA and BS, Computer Science and Mechanical Engineering at University of California, BerkeleyMaster's Degree, Computer Science, Master's Degree, Computer Science at M.I.T.
Github Skills (12)
algorithm10
numerical-optimization10
machine-learning10
c-language10
cprogramming-language10
optimisation10
optimizers10
optimization10
linear-regression10
linear-algebra9
deep-learning8
sparse-matrix8
Programming languages (8)
Github contributions (5)
mlpack/ensmallen
Dec 2014 - Sep 2017
A header-only C++ library for numerical optimization --
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:68 commits in 2 years 8 months
Contributions summary:Stephen's contributions focus on developing a solver for sparse Linear Regression Semi-Definite Programming (LR-SDP) within the numerical optimization library. This involves implementing new functions and classes, specifically `LRSDPFunction`, and integrating them with existing optimization frameworks. Code changes include the implementation of the core evaluation and gradient methods for the objective function, and tests related to it.
optimization-methodsoptimization-algorithmsnumerical-optimizationheader-onlycplusplus
mlpack/mlpack
Dec 2014 - Sep 2017
mlpack: a fast, header-only C++ machine learning library
Role in this project:Back-end Developer
Back-end DeveloperContributions:102 commits, 13 PRs, 15 pushes in 2 years 8 months
Contributions summary:Stephen's contributions focus on implementing a sparse, low-rank Semi-Definite Programming (SDP) solver within the mlpack library. Their work includes the initial development of the solver, as evidenced by commit messages like "WIP: first cut at sparse LR-SDP solver" and associated code modifications in `lrsdp_function.cpp` and `lrsdp_function.hpp`. Furthermore, the user appears to be refining and debugging this solver, as indicated by subsequent commits that rename variables and fix compilation errors, suggesting an iterative development process.
regressionheaderdeep-learningscientific-computingc-plus-plus
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.