AI in production
The infrastructure behind AI systems that actually ship: MCP servers, controlled tool access, LLM integration with real permissions and cost control.
Learn more
Six areas, one person to talk to. From AI in production through process automation to migrating onto AWS.
All of it running in production today, not just on slides.
Most problems that land on my desk are not code problems. They are decisions that were right years ago and now cost money, speed or nerves: an AWS bill nobody can explain any more, a workflow that eats days every month, a system nobody wants to touch.
That is exactly where I work. Always with the same pattern: first understand how the system really behaves, then decide what has to be built, then deliver, in steps that each carry value and can each be turned back.
The infrastructure behind AI systems that actually ship: MCP servers, controlled tool access, LLM integration with real permissions and cost control.
Learn moreReplacing manual workflows with real systems: wired into the software you already run, with permissions and an audit log instead of a chain of tools.
Learn moreI find where your AWS budget leaks away, and cut it measurably without giving up performance or availability.
Learn moreFrom your own data centre or another cloud onto AWS, with a cost model before the move and a rollback path for every step.
Learn moreModernizing grown systems step by step: strangler fig instead of a rewrite, operations untouched, every step reversible.
Learn moreBackends for SaaS and platforms that hold under real load: Go and PHP 8, event-driven, with tenant isolation and recovery designed in.
Learn moreThe same approach across every area. No waterfall, no months without a result. You decide after each phase whether it continues.
You describe the system and the goal. I tell you honestly whether I am the right person. No pitch, no slides.
I read the code, the infrastructure and the incident history. What comes out is a ranked list, not an impression.
Work ships in increments that can go to production individually and be rolled back individually.
Pairing, runbooks and documented decisions. The goal is your team, not an extension of my contract.
It starts with a 30-minute call where we work out what you need and whether I am the right person for it. Then comes a short analysis phase with a concrete deliverable, a risk list, a target architecture or a plan of action, and only then implementation. You decide after each phase whether and how it continues.
Remote from Germany is the default, and has been since 2003. Workshops, kick-offs or architecture sessions on site are possible by arrangement where they make a real difference. Day-to-day work runs asynchronously with fixed times for alignment.
Yes, every service has a deliberately small entry point: a cost analysis, a process survey, a feasibility check or an assessment. Self-contained pieces of work with a clear result in one to four weeks. Not every problem needs a project.
Both, and the larger part is implementation. I write code, build infrastructure and pipelines, and work alongside your team while doing it. In my experience, advice without implementation produces documents nobody reads.
Billing is by day rate, or fixed price for clearly bounded pieces of work. I give you the order of magnitude in the first call, as soon as scope and timeline are clear, without having to check back with an agency, because there is not one in between.
German and English. Several of the projects behind the case studies ran entirely in English, with distributed teams across Europe and the US. Documentation, code and reviews in either language, whichever fits your team.