I'm a machine learning engineer for batteries, living in Germany. I started as an electrician, spent time building high-voltage battery prototypes in automotive, and then got pulled deeper into machine learning for battery systems.
Recently, I’ve been building neural surrogate models for dynamic battery operation. Before that, I consulted for a Silicon Valley battery startup and built the open-source battery-ML project that led to that engagement.
I checked off my travel bucket list early: solo motorbike through Vietnam, and a Work and Travel chapter before that.
Why batteries, why AI
I got into the field by accident after undergrad, joining an 800V/750kW automotive battery project. I stayed because batteries sit at a challenging intersection: electrical, thermal, mechanical, electrochemical, safety, and control constraints all interact, and small improvements matter a lot at battery-industry scale.
That complexity pulled me through hardware, embedded systems, simulation, and eventually deep learning. I see neural nets as a generic mesh and backprop as a generic solver. That framing shapes how I think about battery ML: useful when it cuts down expensive engineering and testing loops, learns from messy operational data, and captures hardware-specific quirks that are hard to model explicitly. Dangerous when validation gets too clean, and most valuable when it behaves sensibly outside idealized data.
What I bring
- Bridging hardware, physics, and ML. I’ve worked close to real battery hardware, simulation, embedded systems, and neural modelling. That helps me connect constraints across disciplines and question assumptions that look fine in isolation but fail at system level.
- System-level trade-offs. Across thesis work, consulting, and independent research, I’ve benchmarked model families under matched hyperparameter budgets and realistic operating profiles. I care less about narrow benchmark wins than about models that are reproducible, robust, and useful inside an engineering workflow.
- Structure in ambiguity. I’ve repeatedly taken open-ended technical problems and turned them into testable progress, from high-voltage prototypes and embedded AI hardware to fast-moving production codebases and neural battery modelling.