Jonas Rauber
Machine Learning Researcher at Apple
Summary
🤩
RockstarJonas Rauber is a Machine Learning Researcher and Engineer with 12 years of experience applying rigorous testing and engineering practices to open-source ML frameworks. Based in Brazil, he has contributed substantive QA, test automation, and bug fixes to high-profile projects like JAX and TensorFlow Probability, improving examples and test coverage for core primitives such as foreach_loop and cross-entropy losses. He also helped evolve adversarial-attack tooling in Foolbox by refactoring code, adding L1-based attacks, and optimizing implementations for robustness research. Comfortable bridging research and production, Jonas blends careful validation with practical model improvements to make ML codebases more reliable and reproducible. An often-overlooked strength is his focus on examples and tests—work that quietly raises the quality bar for downstream users and researchers.
12 years of coding experience
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (14)
testing10
pytorch10
machine-learning10
adversarial-attacks10
tensorflow10
jax10
python10
numpy10
neural-network8
artificial-neural-networks8
probabilistic-programming7
data-science6
ml5
mle5
Programming languages (19)
Github contributions (5)
bethgelab/foolbox
Feb 2020 - Mar 2021
A Python toolbox to create adversarial examples that fool neural networks in PyTorch, TensorFlow, and JAX
Role in this project:
ML Engineer
ML EngineerContributions:53 releases, 5 reviews, 131 commits in 1 year 1 month
Contributions summary:Jonas primarily contributed to the development and improvement of the `foolbox` library, which focuses on creating adversarial examples for machine learning models. Their contributions included refactoring code, improving the attack implementations, and adding new attacks such as L1 FGM, L1 BIM and SparseL1DescentAttack. The user also worked on optimizing the existing code by replacing format strings and using the `ep.norms` library.
pytorchadversarial-attackspythondeep-learningadversarial
jax-ml/jax
Jan 2019 - May 2020
Composable transformations of Python+NumPy programs: differentiate, vectorize, JIT to GPU/TPU, and more
Role in this project:QA Engineer / Test Automation Engineer
QA Engineer / Test Automation EngineerContributions:3 PRs, 43 comments, 17 issues in 1 year 4 months
Contributions summary:Jonas primarily contributed to the testing of the `jax` library. Their commits focused on adding and modifying tests for the `foreach_loop` function, specifically including test cases and fixing existing tests. They also updated examples in the repository, correcting cross-entropy losses within the MNIST examples, improving the overall testing coverage and example functionality.
pytorchpythonjitautomatic-differentiationgpu
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.