Anton Obukhov
Principal Research Scientist at Huawei Switzerland
Zurich, Zurich, Switzerland
Summary
🤩
Rockstar🎓
Top SchoolAnton Obukhov is a Principal Research Scientist focused on computer vision and generative AI, currently developing camera imaging solutions at Huawei Research Center Zürich. With a PhD from ETH Zürich and a decade of industry experience, he has bridged cutting-edge research and product engineering—from contributing to NVIDIA's early CUDA ecosystem to shipping video camera products at Ubiquiti. At ETH he led teams applying generative models to photogrammetry and remote sensing, and his open-source contributions include improving torch-fidelity to make GAN evaluation and TensorFlow-to-PyTorch checkpoint conversion more robust and reproducible. Based in Zurich, he combines deep academic rigor with practical systems experience, often surfacing subtle implementation fixes (e.g., bilinear interpolation integration) that materially improve model fidelity.
10 years of coding experience
14 years of employment as a software developer
Master, Computer Science, Master, Computer Science at Moscow State University, Computational Mathematics and Cybernetics dept.Doctor of Science, Computer Vision, Machine Learning, Doctor of Science, Computer Vision, Machine Learning at ETH Zürich
English, Russian, German
Github Skills (12)
generative-model10
pytorch10
machine-learning10
python10
tensorflow9
repr9
evaluation9
interpolation9
rep9
cgan8
metric8
dcgan8
Programming languages (5)
Github contributions (5)
toshas/torch-fidelity
Apr 2020 - Nov 2021
High-fidelity performance metrics for generative models in PyTorch
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:2 releases, 100 commits, 31 PRs in 1 year 7 months
Contributions summary:Anton primarily worked on enhancing the `torch-fidelity` library for evaluating generative models in PyTorch. They integrated external contributions, added checks, and improved the codebase for scripting. Furthermore, they implemented bilinear interpolation and integrated it into the inception model to address discrepancies. The user also focused on improving reproducibility by adding TensorFlow-to-PyTorch checkpoint conversion tools.
evaluationprecisionreproducibilitygenerative-modelsmetrics
toshas/torch_truncnorm
Oct 2020 - Aug 2021
Truncated Normal Distribution in PyTorch
Contributions:14 commits, 2 PRs, 8 pushes in 10 months
pytorchdeep-learningnormaltruncatednormal-distribution
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.