Who is behind Tippel.
Tippel is an owner-managed AI engineering company from Hombressen, Germany. Lukas Friedrich founded it and leads it: he plans the project, writes the code and puts the system into operation at the client’s site.
Founder & AI Engineer · Hombressen, Germany
From research into production
Physics at the University of Freiburg, focused on deep learning for 3D imaging data, trained on high-performance computing clusters. Research left a mark on every Tippel project: a result only counts once it is reproducible and measured.
Since graduating, Lukas Friedrich has trained neural networks on HPC clusters, built agentic systems on language models and run inference on his own GPU hardware. He founded Tippel so that the same engineer designs a system, builds it and puts it into operation at the client.
This work has produced systems of our own, among them a coding agent and a validation layer for security rules in Microsoft Sentinel. Projects →
| Lead | Lukas Friedrich, Founder & AI Engineer |
|---|---|
| Education | Physics, University of Freiburg |
| Focus | deep learning for 3D imaging data |
| Infrastructure | HPC clusters, our own GPU server |
| Location | Hombressen, Hesse, Germany |
| Languages | German, English |
One engineer, from sketch to handover
What that means for working together and what happens if he is unavailable.
- 01
The advantage
Whoever designs the system also builds it and knows every line. Between requirement and code there is no handover in which knowledge gets lost. Projects that need a large team for many months, we decline.
- 02
The continuity plan
Every project is ready for handover from week one: complete source code, written documentation and reproducible deployment live in your repositories. They rest on widely used tools, so another engineer can take over. Data protection & operations →
How we work
Open, widely used tools instead of homegrown frameworks, so that no system depends on us. Data protection comes from architecture: data flows are documented and follow your requirements, via cloud services or on your own servers. A result only counts once it is measured and reproducible — that holds for the Readiness Check as much as for acceptance.