David Weber
Statistical Developer
Greater Sacramento United States
Summary
👤
Senior🎓
Top SchoolDavid Weber is a Statistical Developer with a Ph.D. in applied mathematics from UC Davis and a decade of experience blending rigorous math with practical software engineering. Currently at Carnegie Mellon’s Delphi group, he builds tools for real-time disease forecasting that translate advanced statistical ideas into production-ready pipelines. His research background includes generalizing the scattering transform to analyze raw sonar wave-fields, reflecting deep signal-processing and applied-ML expertise. As an active contributor to the FluxML ecosystem, he extended Zygote.jl with correct FFT adjoints and adapted Flux.jl activations for automatic differentiation—work that improves gradient-based workflows used by many ML practitioners. Based in Greater Sacramento, he combines academic rigor with open-source impact to solve complex, data-driven problems.
10 years of coding experience
Mathematics, Applied mathematics, Mathematics, Applied mathematics at University of California, DavisBachelor of Science - BS, Mathematics and Computer Science, Bachelor of Science - BS, Mathematics and Computer Science at University of Wisconsin-Madison
Github Skills (17)
fftw10
dft10
machine-learning10
gradient10
automatic-differentiation10
fft10
deep-learning10
flux10
neural-network10
fluxor10
julia10
data-science9
ziggy9
zig9
numerical-analysis9
Programming languages (12)
Github contributions (5)
FluxML/Zygote.jl
Jul 2019 - Mar 2020
21st century AD
Role in this project:
ML Engineer
ML EngineerContributions:18 commits, 2 PRs, 11 comments in 8 months
Contributions summary:David focused on extending the functionality of the Zygote.jl library by adding support for common FFT (Fast Fourier Transform) functions and their adjoints, enabling automatic differentiation capabilities. This involved implementing adjoints for `fft`, `ifft`, `bfft`, `rfft`, `irfft`, `brfft` and related plans, ensuring correct gradient calculations for these operations. The user also contributed to the test suite, adding tests and checks to verify the correctness of the implemented adjoints and the overall behavior of the FFT functions within the Zygote framework.
control-flowautomatic-differentiationmachine-learningjulia-compilerjulia
FluxML/Flux.jl
Sep 2019 - Nov 2019
Relax! Flux is the ML library that doesn't make you tensor
Role in this project:ML Engineer
ML EngineerContributions:17 commits, 1 PR, 7 comments in 2 months
Contributions summary:David's contributions primarily focus on modifying and improving the `activations` function within the Flux.jl library. These changes involved making the `activations` function compatible with Zygote, a library for automatic differentiation, and adapting it to handle empty chains. Further iterations involved refactoring the `activations` implementation using both recursive and iterative approaches. The user's commits also include adjustments to other layers, ensuring their correct functionality and compatibility with the overall framework.
ml-librarythe-human-braindata-sciencedeep-learningneural-networks
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.