Hey, I'm Jan

Senior software engineer at an AI cybersecurity startup. Wrapping up a neuroscience PhD on the side.

I design systems for securing agentic AI. As companies give LLMs and autonomous agents real authority, the problem is governing what those agents can see and do without adding latency they would notice. I architect the control plane that sits inline and makes those decisions in real time, and I own the GPU inference platform it runs on: the layer that has to be fast, reliable, and observable enough to sit in front of everything a company sends to AI.

Before industry, I did electrophysiology across humans, primates, and rodents at the Werner Reichardt Centre for Integrative Neuroscience and the Max Planck Institute for Biological Cybernetics in Tübingen: invasive and non-invasive, MEG, EEG, and MRI, optogenetics and fiber-optic calcium imaging, all in service of decoding noisy, high-dimensional neural data.

Either way, the job is the same: a large, complex system, signal buried in noise, and the discipline to find the one constraint that makes the rest legible. Get the constraint right and the architecture follows from it. The frameworks change, the constraint doesn't.

If you're working on something interesting in AI, security, or GPU inference, get in touch.