Scott Cyphers
Software Engineer at SambaNova Systems
San Diego, California, United States
Summary
🤩
Rockstar🎓
Top SchoolScott Cyphers is a seasoned software engineer with 10+ years focused on deep learning compilers, runtimes, and performance engineering, currently building systems at SambaNova in San Diego. He helped originate and evolve nGraph at Nervana/Intel and has driven ONNX/Glow-based inference compilation and accelerator support at Lightmatter and Intel, blending compiler internals with practical deployment needs. His open-source contributions to high-profile projects like nGraph and Microsoft CNTK emphasize performance optimization, memory allocator improvements, and robustness fixes that often address subtle correctness and serialization edge cases. With an MIT background and decades of systems-level experience dating back to Lisp-era processors and databases, he brings rare depth across low-level runtime, autodiff/compiler design, and production ML toolchains. An understated strength is his ability to refactor complex compiler/runtime components for both maintainability and measurable runtime gains.
10 years of coding experience
43 years of employment as a software developer
SM Electronics Computers and Systems, SM Electronics Computers and Systems at Massachusetts Institute of Technology
Github Skills (19)
c-language10
runtimes10
run-time10
cntk10
deep-neural-networks10
optimisation10
cplus10
cpp10
cprogramming-language10
optimization10
data-structure9
serializer9
serializable9
serialization9
machine-learning9
Programming languages (6)
Github contributions (5)
NervanaSystems/ngraph
Aug 2017 - Oct 2020
nGraph - open source C++ library, compiler and runtime for Deep Learning
Role in this project:
Back-end Developer & Performance Engineer
Back-end Developer & Performance EngineerContributions:98 releases, 1 review, 653 commits in 3 years 2 months
Contributions summary:Scott focused on performance optimization and functionality within the nGraph deep learning compiler and runtime. They contributed to the serialization and deserialization of node inputs, specifically addressing issues with non-zero output indices. The user also implemented new features for dynamic shapes and made efforts in performance improvements to the memory allocator functionality. The user also refactored various components of the compiler and runtime.
inference-enginecppc-librarydeep-learningtvm
microsoft/CNTK
Jul 2015 - Sep 2015
Microsoft Cognitive Toolkit (CNTK), an open source deep-learning toolkit
Role in this project:Back-end Developer
Back-end DeveloperContributions:62 commits in 2 months
Contributions summary:Scott primarily focused on code modifications within the `DataReader` and `MachineLearning` directories of the CNTK project. Their contributions included bug fixes related to memory leaks, format string adjustments, and addressing compilation warnings. These changes indicate a focus on improving the stability and maintainability of the codebase, specifically within the data reading and machine learning components. The user also made changes to improve the compilation process.
pytorchpythondeep-learningc-plus-plusmachine-learning
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.