I spent my PhD teaching neural networks to see past the diffraction limit. The methods carried over to enterprise data better than I expected.
Three areas, one habit: understand the system before changing it.
Algorithms for biological image analysis, super-resolution microscopy, and automated quality control with nanometer precision.
Neural network architectures that build domain knowledge into the model instead of hoping the data supplies it.
Coding agents that write and test ABAP. RAG over SAP data on AI Core. Deployed inside existing landscapes, not beside them.
Where the methods came from and where they went.
University of Würzburg - Developed AI algorithms for super-resolution microscopy, combining compressed sensing with deep neural networks.
Coding agents that write and test ABAP on BW/4HANA. RAG over SAP data on SAP AI Core. LangGraph agents that run inside existing landscapes.
Independent development of a full-stack AI platform for sports analytics. Live at endurexai.de.
What I am working on now.
Transformers for Scientific Imaging
Investigating the application of transformer architectures to gain additional positional information over large timeframes in single-molecule localization microscopy datasets.
Status: Preprint Published
Multi-Institutional Research
Ongoing collaborations with research institutions on AI applications in biological imaging, neuroscience, and medical diagnostics.
Super-resolution microscopy and AI integration
Cross-institutional AI methodology development
Enterprise AI implementation consulting
How I work, in four steps.
Write the math down before the code.
Build it to run fast: modern frameworks and GPU acceleration.
Test on real datasets and report the numbers, including the bad ones.
Move what works into production systems.
The numbers.
If you have a problem that is difficult for the right reasons, I want to hear about it.