What I'm doing now

Current focus

My work centres on AI inference orchestration and cost optimization for ML and large language models — most recently leading teams on this at Amazon in Berlin. Right now I'm going deeper on two questions that have occupied me for a while: how do you make production AI economically viable at scale, and how do you architect agentic systems that stay reliable as the patterns mature? Both are harder than they look from the outside.

Alongside that I'm building out an open-source portfolio — an inference simulator, an agent evaluation harness, a cloud-native service on Kubernetes — and working through certifications in agentic AI (NVIDIA NCP-AAI, Anthropic CCA Foundations).

Writing about

• The economics of AI inference at scale — why most cost-optimization advice misses the architectural wins.
• The taxonomy of agents — distinguishing workflows, tool-using calls, single agents, and multi-agent systems.
• How engineering organizations adapt (or don't) to AI as a first-class part of the stack.

Reading

• Designing Data-Intensive Applications — Martin Kleppmann (a re-read; it ages well).
• The Manager's Path — Camille Fournier.
• Recent inference-optimization papers from Anthropic, Together AI, and the vLLM project.

Open to conversations about

Engineering leadership in AI-native companies, production AI architecture and inference economics, and cross-border engineering teams. If any of that overlaps with what you're working on, I'd be glad to hear from you.

Based in

Berlin, Germany. Regularly in conversations across European and US time zones.