Андрей Стоцкий (RuRo)
Research Scientist at Tevian LLC
Summary
🤩
RockstarАндрей Стоцкий (RuRo) is a research scientist and CV/ML researcher with 11 years of hands‑on experience improving machine learning frameworks, tooling, and test infrastructure. He teaches computer vision courses at leading Russian institutions (MSU, YSDA, MIPT, HSE) while contributing to major open‑source projects—improving numerical stability and ONNX export in Apache MXNet and strengthening test suites in setuptools and Sphinx. His work blends rigorous QA/test automation with backend and ML engineering, evidenced by fixes for softrelu overflow, symlink handling, and comprehensive test coverage. Colleagues describe him as competent, hardworking and modest; he pairs deep technical competence with a preference for practical, reliability-focused contributions that quietly elevate widely used developer tools.
11 years of coding experience
Stackoverflow
343reputation
240kreached
6answers
10questions
Github Skills (37)
restructuredtext10
pytest10
python10
sphinx10
setuptools10
testing10
mxnet10
machine-learning10
data-export10
onnx10
exporter10
deep-learning10
exports10
documentation10
test-automation10
Programming languages (22)
Github contributions (5)
pypa/setuptools
Jan 2020 - Aug 2023
Official project repository for the Setuptools build system
Role in this project:
QA Engineer / Test Automation Engineer
QA Engineer / Test Automation EngineerContributions:2 reviews, 1 PR, 14 comments in 3 years 7 months
Contributions summary:Андрей primarily focused on enhancing the testing framework and ensuring the quality of the `setuptools` project. They added new tests for sdist handling, including testing for files listed in Extension.sources and Extension.depends, and improved the existing test suite. They also refactored test code and corrected handling of symlinked extension sources, contributing significantly to test coverage and reliability.
setuptoolsbuild-system
apache/mxnet
Mar 2020 - Jun 2020
Lightweight, Portable, Flexible Distributed/Mobile Deep Learning with Dynamic, Mutation-aware Dataflow Dep Scheduler; for Python, R, Julia, Scala, Go, Javascript and more
Role in this project:ML Engineer
ML EngineerContributions:5 commits, 7 PRs, 71 comments in 3 months
Contributions summary:Андрей primarily contributed to improving the numerical stability of fused operators within the MXNet framework. They addressed potential overflow issues in the `softrelu` activation function by implementing a fix and a corresponding test. Furthermore, the user worked on adapting the library for ONNX export, specifically implementing translations and test cases for features such as LSTM, `slice_axis`, and `topk` operations, including implementing helper functions for reshaping and tensor creation. These changes enhanced the interoperability of MXNet models.
pythonschedulerdataflowmutationdata-science
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.