Cyrus Asfa
Research at KLAY
London, England, United Kingdom
Summary
👤
Senior🎓
Top SchoolCyrus Asfa is a research-focused software engineer with 11 years' experience at the intersection of machine learning and music, currently leading generative music efforts as a founding engineer at KLAY in London. He holds a PhD-era background in perceptual audio and deep learning from Queen Mary’s C4DM and has built waveform-based, real-time music generation systems during a stint as a Research Scientist at ByteDance. Cyrus founded Sonophase AI to tackle music search and discovery, and has collaborated with CNRS on differentiable optimisation and wavelet scattering for music analysis. An active open-source contributor, he’s improved core functionality and backend compatibility in the popular kymatio scattering library, bringing practical GPU-accelerated signal-processing expertise to ML stacks. Known for blending research rigor with product-minded engineering, he combines signal-processing depth with scalable generative system design.
11 years of coding experience
4 years of employment as a software developer
BEng, Computing, BEng, Computing at Imperial College LondonMSc, Sound and Music Computing, Distinction, MSc, Sound and Music Computing, Distinction at Queen Mary University of London
Github Skills (10)
pytorch10
refactoring10
tensorflow10
backend10
back-end-development10
python10
wavelet10
numpy10
pytest9
testing9
Programming languages (8)
Github contributions (5)
kymatio/kymatio
Jun 2022 - Jan 2023
Wavelet scattering transforms in Python with GPU acceleration
Role in this project:
Back-end Developer
Back-end DeveloperContributions:153 reviews, 14 commits, 45 PRs in 7 months
Contributions summary:Cyrus primarily contributed to refactoring and improving the core functionality of the kymatio library, focusing on deprecating and replacing the `average` parameter within the `Scattering1D` module. They addressed bugs, updated tests, and introduced new semantics for parameters like `T`. Additionally, they worked on reshaping input and output data for different backends like NumPy, TensorFlow and PyTorch, and implemented core functionalities within the 1D base frontend and other frontend implementations. These changes involved core library improvements, and refactoring efforts aimed at improving code maintainability and usability.
pytorchwaveletswaveletpythongpu-acceleration
adamhosier/WACC-compiler
Oct 2015 - Apr 2020
Contributions:76 pushes, 2 branches in 4 years 6 months
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.