Scott Cheng

State College, Pennsylvania, United States

Join Prog.AI to see contacts

Summary

🤩

Rockstar

🎓

Top School

Scott Cheng is a deep learning and high-performance computing researcher with a decade of software engineering experience, currently pursuing doctoral research at Penn State focused on optimizing and modeling the performance of LLMs, DLRMs, and AlphaZero-style algorithms on supercomputing systems. He blends rigorous academic training from top Taiwanese universities with hands-on contributions to major open-source projects like PyTorch, where he optimized CUDA kernels and cross-platform support, and front-end improvements to interactive demos in community repositories. Scott’s work sits at the intersection of systems performance and ML model scaling, delivering practical kernel-level improvements as well as usability-focused UI fixes. Based in State College, PA, he brings both deep algorithmic insight and attention to implementation detail—often spotting platform-specific bottlenecks that others miss.

10 years of coding experience

Master's degree, Computer Science, 4.22 / 4.30, Master's degree, Computer Science, 4.22 / 4.30 at National Chiao Tung University

Doctor's Degree, Computer Science, 3.85 / 4.00, Doctor's Degree, Computer Science, 3.85 / 4.00 at Penn State University

Bachelor's degree, Concurrent/Dual degrees: Mathematics, Computer Science, 4.05 / 4.30, Bachelor's degree, Concurrent/Dual degrees: Mathematics, Computer Science, 4.05 / 4.30 at National Tsing Hua University

Github Skills (16)

html10

cuda10

pytorch10

machine-learning10

javascript10

gpu10

python10

cprogramming-language9

canvas9

neural-network9

c-language9

deep-learning9

css9

tensor8

autograd8

Programming languages (10)

C#C++JavaScriptGoPHPObjective-CHTMLMarkdown

Github contributions (5)

ricktu288/ray-optics

May 2016 - Sep 2016

A web app for creating and simulating 2D geometric optical scenes, with a gallery of (interactive) demos.

Role in this project:

Front-end Developer

Contributions:11 commits, 1 PR, 9 pushes in 3 months

Contributions summary:Scott primarily focused on front-end development, as evidenced by the HTML and JavaScript modifications. Their contributions included linting and code cleanup in JavaScript files, along with improvements to the HTML structure and styling within `light.html`, `index.html`, and `simulator/index.html`. The changes also involved optimizing images and updating the meta tags.

opticsoptics-simulationcanvashostedsimulation

pytorch/pytorch

Jul 2021 - Dec 2022

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Role in this project:

ML Engineer

Contributions:1 review, 12 commits, 8 PRs in 1 year 4 months

Contributions summary:Scott contributed to the PyTorch repository by improving CUDA kernels for distance calculations and normalization functions. Their work included refactoring existing code using `BlockReduce` for efficiency, optimizing performance, and adding comments for clarity. They also addressed platform-specific path issues on Windows, improving the cross-platform compatibility of the library. Additionally, the user improved Intel compiler support.

pythongpu-accelerationdeep-learninggpunumpy

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.

Request Free Trial