Matthias Seeger
Principal Applied Scientist (Machine Learning) at Amazon
Berlin, Germany
Summary
🤩
Rockstar🎓
Top SchoolMatthias Seeger is a Principal Applied Scientist in Machine Learning based in Berlin with 12+ years of experience building production-grade ML systems and AutoML features for Amazon SageMaker. He blends deep academic roots—from a PhD in Informatics and faculty roles at EPFL and Saarland—with hands-on engineering contributions to major open-source projects like MXNet and the d2l deep-learning book used at top universities. At Amazon he focuses on Automatic ML, bringing algorithmic rigor to scalable hyperparameter optimization and scheduler design (notably improvements to ASHA/Hyperband logic). His work spans low-level numerical operators to high-level AutoML workflows, reflecting an unusual breadth across research, backend systems, and ML infrastructure.
12 years of coding experience
10 years of employment as a software developer
Master of Science (MSc) (German: Diplomingenieur), Computer Science, Master of Science (MSc) (German: Diplomingenieur), Computer Science at Karlsruhe Institute of TechnologyDoctor of Philosophy (PhD), Informatics, Doctor of Philosophy (PhD), Informatics at The University of Edinburgh
English, German, French
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (24)
pytorch10
c-language10
python10
mxnet10
machine-learning10
ensemble-learning10
scheduler10
hyperparameter-optimization10
deep-learning10
tensorflow10
cprogramming-language10
linear-algebra10
numerical-computing10
storybook9
textbook9
Programming languages (3)
Github contributions (5)
autogluon/autogluon
Oct 2019 - Oct 2021
Fast and Accurate ML in 3 Lines of Code
Role in this project:
ML Engineer
ML EngineerContributions:71 reviews, 22 commits, 48 PRs in 2 years
Contributions summary:Matthias contributed to the core functionality of the AutoGluon library, focusing on the development and optimization of Hyperband schedulers, including the promotion and stopping logic for the asynchronous ASHA variant. Their work involved refactoring and extending scheduler components, integrating improvements for model-based searchers, and addressing issues within the Hyperband scheduler. Additionally, they implemented changes to improve the performance and overall stability of the hyperparameter optimization process.
forecastingimage-textmlppythonmeta-learning
d2l-ai/d2l-en
Oct 2021 - Dec 2022
Interactive deep learning book with multi-framework code, math, and discussions. Adopted at 500 universities from 70 countries including Stanford, MIT, Harvard, and Cambridge.
Role in this project:ML Engineer
ML EngineerContributions:5 reviews, 87 commits, 7 PRs in 1 year 1 month
Contributions summary:Matthias contributed to the development and refinement of machine learning components and hyperparameter optimization within the deep learning book repository. Their work involved integrating various machine learning frameworks like PyTorch, TensorFlow, and MXNet, and implementing hyperparameter tuning methods. The commits showcase the user's effort in building and improving the library's infrastructure, fixing existing issues, and refactoring code to enhance its functionality.
kagglepythontensorflowgaussian-processesnatural-language-processing
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.