About
Imaging scientist, Apple-platform developer, and independent AI researcher. Building instruments that make complex systems easier to see, measure, and remember.
In my career spanning over three decades, I have been working at the intersection of life sciences, imaging, and software engineering. Lately I started working with artificial intelligence. My focus is two fold: how can I improve the transformer architecture and what tools can enhance the capabilities of AI. This translates to two current projects. Apertura is a transformer architecture where every processing step and every buffer is fully observable. ES Memory is an MCP tool that lets AI remember. The fundamental question that has consistently guided my work? What insights can instruments provide. What can we observe, image, and improve.
It started in neuroscience and cell biology, at the University of Würzburg under Martin Heisenberg. How do genes, brain structure, and behavior relate was the overarching question. Later at the University of Oregon with Andrew Bajer, we elucidated the mechanics of cell division, the mitotic spindle and the dynamics of microtubles. The common theme was building instruments that turned observation into quantitative data.
Then a transition into industry: six years at Leica Microsystems. First in factory in Heidelberg and in the U.S. offices. I transitioned from writing the application software for confocal microscopes to applying software solutions to practical imaging problems for clients like 3M and Colgate.
I moved back to academia, directing the Light Microscopy Core at Cedars-Sinai Medical Center for eighteen years. Running a core lab is never about one’s own science, but supporting research by making sure the scientist get the best possible results. I took an active role, enhancing experimental designs, adapting imaging workflows and building custom analytical tools. Is was exciting working across such diverse areas of research such as endocrinology, oncology, neuroscience, and virology.
Since 2020 I work independently on memory architecture on local-first AI systems. I believe true intelligence in AI doesn’t emerge from a model alone, but it emerges from the recursive interaction with memory. Most AI memory systems built today treat memory as a filing cabinet. I treat AI memory more like a personal filing cabinet. And I still write code in Objective-C. Intentionally
I enjoyed education in the German humanist tradition of Bildung. It still shapes how I think: serious questions deserve depth and not just polished answers. A longer account of all this lives at the full biography.