Fabian Isensee
Group Lead at DKFZ German Cancer Research Center
Kirchardt, Baden-Württemberg, Germany
Summary
🤩
Rockstar🎓
Top SchoolFabian Isensee is a Group Lead at the DKFZ with 11 years of experience combining deep learning research and hands-on engineering in medical image analysis. He holds a doctorate from the University of Heidelberg and a strong molecular biotech background, which helps him bridge domain science and ML model development. An active open-source contributor, he has extended flagship projects such as nnU-Net and MITK—adding novel loss functions, mixed-precision training, and a median statistic for medical imaging—while also improving QA and inference pipelines. Fabian blends full-stack development experience from ilastik with rigorous test automation practices, making him as comfortable shipping robust production code as prototyping new segmentation methods. Colleagues rely on him for turning complex biomedical requirements into reproducible, well-tested tooling that accelerates clinical research.
11 years of coding experience
Dr. rer. nat., Dr. rer. nat. at University of HeidelbergMaster of Science (M.Sc.), Molecular Biotechnology, 1.2, Master of Science (M.Sc.), Molecular Biotechnology, 1.2 at Heidelberg University
English, French, German
Stackoverflow
1reputation
164reached
2answers
1question
Github Skills (24)
pytorch10
c-language10
pyq10
il10
loss-functions10
python10
image-processing10
testing10
machine-learning10
user-interface10
graphical-user-interface10
human-interface10
image-segmentation10
pyqt10
segmentation10
Programming languages (7)
Github contributions (5)
MIC-DKFZ/nnUNet
Apr 2019 - Sep 2022
Role in this project:
ML Engineer
ML EngineerContributions:3 releases, 6 reviews, 688 commits in 3 years 5 months
Contributions summary:Fabian's commits primarily focus on improving and extending the nnU-Net framework for medical image segmentation. This is demonstrated through the implementation of various loss functions, including Focal Loss and GDL, as well as enhancements to the data augmentation pipeline and the addition of new trainers, such as the PreAct Residual UNet and associated training configurations. The user also made improvements to the inference capabilities, including incorporating mixed-precision training and various postprocessing steps.
segmentation
ilastik/ilastik
Oct 2015 - Jun 2016
ilastik-shell, applets, and workflows to string them together.
Role in this project:Full-stack Developer
Full-stack DeveloperContributions:11 commits, 1 PR, 4 pushes in 8 months
Contributions summary:Fabian primarily worked on the `FeatureSelectionDialog` within the `ilastik` project. They implemented the UI elements, including the combo box, buttons, and text descriptions, for the feature selection tool. The user also addressed issues related to axis tag problems and made improvements to the feature selection process, including adding the ability to show feature names. Additionally, they attempted to integrate `opSimplePixelClassification`.
to-stringpythonstringilastikmachine-learning
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.