Peter Goldsborough
Co-Founder And CTO at Marine Corps Cyber Auxiliary
Washington, District of Columbia, United States
Summary
🤩
Rockstar🎓
Top SchoolPeter Goldsborough is a Co-Founder and CTO in Washington, D.C. with 11 years of experience building low-level ML systems, distributed infrastructure, and defense-focused software. An early PyTorch/C++ contributor and former Facebook/FAIR engineer, he brings deep C++/CUDA and compiler expertise—contributing to PyTorch, Glow, Caffe2 and scikit-image—and is known for pragmatic low-level fixes, from rewriting legacy C into modern C++ for audio processing to correcting CUDA LLTM kernels. At Anduril he led software for expeditionary data centers, Army C2 and JADC2 initiatives, and now applies that systems-to-product experience to logistics at Rune Technologies. He also serves in the Marine Corps Cyber Auxiliary, blending operational cyber advising with hands-on engineering, and holds a BSc in Computer Science from Technical University Munich.
11 years of coding experience
10 years of employment as a software developer
Bachelor of Science (B.Sc.), Computer Science, Bachelor of Science (B.Sc.), Computer Science at Technical University MunichBG/BRG St.Martin Villach
German, English, French, nerd
Stackoverflow
1,388reputation
181kreached
22answers
12questions
tensorflow
top-5%
Github Skills (60)
torchscript10
unit-testing10
code-optimization10
pytorch10
c-language10
caffe10
documentations10
pytest10
python10
operation10
image-processing10
testing10
tensorrt10
machine-learning10
audio-processing10
Programming languages (15)
Github contributions (5)
pytorch/extension-cpp
Mar 2018 - Aug 2018
C++ extensions in PyTorch
Role in this project:
ML Engineer
ML EngineerContributions:19 commits, 1 PR, 13 pushes in 5 months
Contributions summary:Peter contributed to the PyTorch C++ extension repository by fixing CUDA kernels to support batch sizes greater than one. They added and refined checks for forward and backward passes, ensuring the correctness of the LLTM implementation. Furthermore, they updated the codebase by making CUDA tensors contiguous and correcting the bias dimension. These changes demonstrate a focus on improving the performance and accuracy of the CUDA-based LLTM implementation within the PyTorch ecosystem.
pytorchcudacpplibtorchextensions
scikit-image/scikit-image
Aug 2017 - Nov 2017
Image processing in Python
Role in this project:Data Scientist
Data ScientistContributions:20 commits, 1 PR, 18 comments in 2 months
Contributions summary:Peter primarily contributed to the development of a `generate_shapes` module within the scikit-image library. Their work involved the implementation of functions to create various geometric shapes (rectangles, circles, and triangles) with random properties, including size and color. These contributions included API redesign, fixing bugs related to integer division in Python 2, and the addition of unit tests, demonstrating a focus on both the functionality and quality of the image generation capabilities. The user also refactored the code to improve efficiency and code readability.
image-processingpythoncomputer-visionimage
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.