Jan Krepl
Senior Machine Learning Engineer at Open Brain Institute
Geneva, Geneva, Switzerland
Summary
👤
Senior🎓
Top SchoolJan Krepl is a Senior Machine Learning Engineer based in Geneva with a decade of experience bridging quantitative finance and large-scale ML research. He has led ML teams and delivered production-ready systems at institutions like EPFL’s Blue Brain Project and Open Brain Institute, while also consulting through Toptal. Jan’s background in quantitative finance (ETH Zürich) and economics gives him a strong grounding in probabilistic modeling and risk-aware solutions, reflected in his open-source work on deepdow for deep-learning portfolio optimization. He focuses on building robust training infrastructure—early stopping, checkpointing, MLflow logging—and rigorous evaluation pipelines that move models from research to reproducible production. Colleagues describe him as a pragmatic problem-solver who combines academic rigor with hands-on engineering across neuroscience, finance, and industry projects.
10 years of coding experience
7 years of employment as a software developer
Master’s Degree Quantitative Finance, Master’s Degree Quantitative Finance at ETH ZürichBachelor's degree Economics, Bachelor's degree Economics at Charles University
Stackoverflow
4,060reputation
322kreached
118answers
0questions
tensorflow
top-5%
scikit-learn
top-1%
machine-learning
top-5%
nlp
top-5%
python
top-5%
Github Skills (23)
pytorch10
python10
evaluation10
machine-learning10
deeplearning-ai10
deep-learning10
scikit-learn10
metric10
mlflow9
convex-optimization9
tensorflow9
tensorboard9
nlp9
pytest8
tfidf-vectorizer6
Programming languages (9)
Github contributions (5)
jankrepl/deepdow
Feb 2020 - Aug 2022
Portfolio optimization with deep learning.
Role in this project:
ML Engineer & Data Scientist
ML Engineer & Data ScientistContributions:1 release, 7 reviews, 54 commits in 2 years 6 months
Contributions summary:Jan primarily contributed to the implementation of core machine learning components for portfolio optimization. They added benchmark models, including one-over-N, random, and singleton strategies, and implemented evaluation logic to assess model performance. Furthermore, they introduced the main framework, including callbacks for early stopping, model checkpointing, and MLFlow logging. These changes suggest a focus on building the core infrastructure for training, evaluating, and comparing deep learning models in the context of portfolio optimization.
pytorchmarkowitzconvex-optimizationwealth-managementdeep-learning
jankrepl/pychubby
Jul 2019 - Sep 2019
Automated face warping tool.
Contributions:97 commits, 5 PRs, 104 pushes in 2 months
transformationfacial-featurespythonlandmark-detectionface-recognition
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.