On Human–Machine Collaboration
01HUMAN-MACHINE-
HUMAN-MACHINE-
HYBRID
Machines do some tasks better. Humans do others. On the tasks that matter, the hybrid team beats either alone.
02HUMAN-MACHINE-
HUMAN-MACHINE-
INTERFACE
No one fully understands a modern AI model — and that is fine. What matters is how we use it: the inputs, the context, the layers around the model.
03HUMAN-MACHINE-
HUMAN-MACHINE-
ADVANTAGE
Find the problem first, then ask whether AI fits. Measure the advantage as return on investment; if it isn't there, AI isn't the answer.
Selected Work
- 1'000× faster
- ML pipeline for UAV aerodynamic surrogates — over 1'000× CFD throughput for rapid prototyping. Automated CAD generation included.
- 90% time
- Agentic AI for social-media automation — cybersecurity consultancy. Daily workload reduced from one hour to five minutes.
- Weeks → hours
- Automated OCR for handwritten visitor letters — Swiss cultural institution and two universities. ~70'000 letters, multiple languages, circular script. Pro bono.
About
Award-winning AI researcher, engineer and consultant based in Zürich. Dr.-Ing. (PhD) in computational mechanics and machine learning; postdoc at ETH Zürich. Applied work spans LLMs, agentic AI systems and AI-driven engineering simulation.
Read full bio