Georg Ostrovski
Research Engineer at Amfileon
London, England, United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolGeorg Ostrovski is a Research Engineer based in London with 12 years of experience bridging rigorous mathematics and production-quality machine learning systems. With a PhD and MSc in Mathematics from the University of Warwick, he has been at DeepMind since 2014, contributing to research engineering while also advising startups as a scientific advisor. Georg has a strong track record in QA and test automation for prominent open-source ML tooling (notably improving robustness in torch/torch7 and torch/nn), where he removed flaky tests, tightened numerical tolerances, and added coverage for new tensor operations. His background includes C++ software development and editorial experience, giving him a rare combination of precise technical implementation skills and clear communication. Colleagues value his methodical approach to reliability and his habit of catching subtle edge cases that often elude standard test suites.
12 years of coding experience
4 years of employment as a software developer
Pre-diploma (Vordiplom), Mathematics, 1.1, Pre-diploma (Vordiplom), Mathematics, 1.1 at Universität DortmundDoctor of Philosophy (PhD), Mathematics, Doctor of Philosophy (PhD), Mathematics at University of Warwick
English, German, Russian, French
Github Skills (8)
lua10
pytorch10
test-automation10
testing10
numerical-computing9
neural-network9
machine-learning8
linear-algebra8
Programming languages (6)
Github contributions (5)
torch/torch7
Jun 2014 - Feb 2016
http://torch.ch
Role in this project:
QA Engineer / Test Automation Engineer
QA Engineer / Test Automation EngineerContributions:26 commits, 28 PRs, 14 comments in 1 year 8 months
Contributions summary:Georg primarily contributed to the quality assurance of the `torch7` repository, focusing on identifying and resolving issues within the test suite. Their work involved removing flaky tests and adjusting precision thresholds for existing tests, ensuring the accuracy and reliability of the library's functionalities. Furthermore, the user added new test cases and functions to validate the newly added `torch.round`, `torch.cpow`, and `torch.mod` functions and the functionality of existing ones, demonstrating a commitment to comprehensive testing coverage. The user also improved test suite reliability by fixing issues in existing tests such as empty tensor handling.
libtorchtorchc-plus-plustensorautograd
torch/nn
Jun 2014 - Apr 2016
Role in this project:QA Engineer / Test Automation Engineer
QA Engineer / Test Automation EngineerContributions:9 commits, 5 PRs, 3 comments in 1 year 9 months
Contributions summary:Georg primarily focused on improving the quality and reliability of the `torch/nn` library's tests. Their contributions included adjusting error tolerances to reduce test flakiness, correcting test implementations, and modifying test input data to prevent failures. These changes were applied across several modules, including `LogSoftmax`, `BCECriterion`, and various tests related to different neural network layers, thereby improving the robustness and stability of the testing framework.
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.