One engineer. Full responsibility.
Tippel is the AI engineering company of Lukas Friedrich. Working with Tippel means working with me: I plan your project, write the code, put the system into operation at your site — and pick up the phone when something needs sorting.
Founder & AI Engineer · Hofgeismar, Germany
Background
Physics at the University of Freiburg, M. Sc. in 2019 — focused on deep learning for 3D imaging data, trained on high-performance computing clusters. Research left a mark on every project I run today: a result only counts once it is reproducible and measured.
Since then I have built the stack bottom-up: neural networks on HPC clusters, language-model systems with several cooperating agents, server infrastructure — and the applications on top that people actually work with. I founded Tippel because mid-sized companies deserve production-grade AI without paying for agency overhead and friction.
Three systems from this work run in production today: KomplAI verifies compliance statements against the official legal texts, Foliovo generates complete websites for paying customers, and the Sentinel validation layer validates security detections for Bluerope Consult GmbH.
Why one person — and what if he is unavailable?
The advantage
No junior developers learning on your budget. No game of telephone between sales, project management and engineering. The person who scopes your project writes the code — and the person who writes the code answers your call. What doesn’t fit into one engineer’s hands, I decline. That is part of the deal.
The continuity plan
I ask myself the fair question too: every project is handed over with complete source code, written documentation and reproducible deployment — in your repositories. From week one you hold working software any competent engineer could take over. No black box, no lock-in. Details for your procurement team →
How I work
Open, widely used tools instead of homegrown frameworks — so you never depend on me. GDPR by architecture, not by disclaimer: data flows are documented and built to your requirements, whether via cloud services or on your own servers. I deliver working software every week, and in the Readiness Check you also get a “don’t build this” when that is the right answer.
Toolbox — for the IT team
- AI agents / LangGraph
- RAG & knowledge search
- LLM fine-tuning
- PyTorch
- vLLM
- Kubernetes
- CUDA / HPC
- FastAPI
- React
- Azure / EU cloud
Let’s find out whether your idea holds
30 minutes are enough for a first honest read — then you decide whether the Readiness Check is the next step.