About Marcel
From physics research to building reliable AI systems
My path started in experimental physics, where precision and uncertainty live side by side. Over time, I became more interested in how those principles apply to real-world systems: how to build technology that knows what it doesn’t know.
The Journey
Academic Foundation
I completed my PhD at Heidelberg University, focusing on light field reconstruction and 3D geometry algorithms. My work explored how to recover the shape, orientation, and material properties of reflective objects, where classical computer vision breaks down.
Physics trained me to deal with uncertainty methodically. Every measurement carries limits, and understanding those limits often matters more than the value itself.
Industry Transition
In 2017, I joined AGT International as a data scientist while finishing my doctorate. That period bridged two worlds: scientific rigor and industrial reality. It showed me how analytical thinking could translate into systems that actually make decisions.
After earning my PhD in 2018, I continued refining that connection, building models that needed to work outside the lab, with all the noise and ambiguity that entails.
rabbitAI (2019–2026)
In 2019, I co-founded rabbitAI to tackle one of the core challenges in AI: reliable ground truth. For seven years we designed systems that capture the real world precisely enough for machine learning to trust it, from automotive cabins to industrial environments.
As CTO, I connected hardware, physics, and machine learning. The aim wasn’t perfection but transparency: AI systems that make their assumptions visible and stay quiet when the data speaks in uncertain terms.
Wayve (since 2026)
Since 2026 I work at Wayve on the data and validation side of autonomous driving: pipelines, data quality, and the tooling that tells us when ground truth is wrong. After seven years of running a company around that problem, I now get to work on it directly.