th3chris
Back

$ kubectl describe principles

Principles

  • Architecture before code: I see every project as a whole. Decisions made early secure scalability, security and maintainability.
  • Scalability is non-negotiable: Good software must scale. Whether ten or ten thousand users — architecture, deployment and data flows have to be built for it.
  • Tests as foundation: "What is not tested does not work." Only with consistent integration tests, realistic scenarios and automated pipelines can software stay stable under load and in production. Happy to work TDD-first: write the test, then the implementation.
  • Realism over lab conditions: Dev environments must mirror production — smaller but real. Only then can errors be caught early. And it must always be possible to reproduce production issues locally.
  • Automation & consistency: A deployment has to run with one command, a test environment with one click. GitHub or Azure Pipelines orchestrate container dependencies, key vault secrets and build chains — automated, for example via Cake.
  • Use AI deliberately: AI can accelerate development tremendously — but an AI feature is only valuable when it is understood, reviewed and cleanly integrated. I use AI as a tool, not as a black box.
  • Transparency & knowledge transfer: I do not leave a black box behind. I leave a solution teams understand and can evolve on their own.

An example from my work

On a client engagement, tests had long been run against simplified in-memory databases while production used Postgres. The result: hours spent debugging, endless log reading, because bugs only surfaced in live operation. Some queries behaved very differently locally — primarily because Entity Framework treats in-memory databases very differently from a real one.

My approach: Containerized integration tests with TestContainers. Instead of artificial test databases, every environment — from the developer laptop to the CI/CD pipeline — now runs real instances of the same database as production. TestContainers.

The outcome:
  • Realistic tests that surface the same bugs that production would later show.
  • No risk to production data, because every instance runs in isolation.
  • Less time in debugging, because problems appear early and reproducibly.

What used to be a permanent pain point became a reliable foundation: developers and teams can trust their tests — and nasty production surprises belong to the past.

Why this matters

Above all, my way of working creates one thing for clients: confidence.

  • Scalability means: Your system grows with your business — without being reinvented every time.
  • Realistic tests means: Bugs are found before they hit you in front of a customer or in production.
  • Automation means: Fewer manual errors, faster cycles and more reliable deployments.
  • Sustainable AI integration means: New features shipped faster — but in a way your team can understand and operate long-term.

In short: You avoid nasty surprises, save time and cost — and get systems you can actually rely on.

Work model — flexible to your needs

Solo Development

Complete solutions from one source

From architecture through implementation to deployment — one point of contact for everything. No overhead, short decision paths, maximum speed.

Quickmail Systems (Montabaur): 4 years of complete mobile app development as a solo developer for Quickmail & Quickpac in Switzerland. 6 independent solutions. Offline-first for mountain regions.

Team Integration

Seamlessly in your existing team

Code reviews, mentoring, joint architecture decisions. I integrate into your team as if I've been there from day one — bringing experience that moves teams forward.

Hoffmann Group: 5+ years in a 5-person Digital Services team. Together built 3 central DataHubs and maintained 8 shared libraries.

Enterprise Architecture

Mastering complex system landscapes

When multiple divisions, legacy systems, and international teams need to work together — that's where I thrive. SAP integration, multi-system orchestration, regulatory compliance.

Schneider Electric: Integration of diverse divisions into a seamless IDE — after Microsoft professionals deemed it "impossible".

AI is my tool — not my autopilot

My approach to responsible use of AI in software development

Every line is understood

AI generates code faster than ever before. But code that isn't understood is technical debt from day one. I use AI for speed — and review, understand, and optimize every result.

Security is not an afterthought

AI-generated code has blind spots: injection vulnerabilities, missing input validation, unchecked edge cases. With 25+ years of experience, I catch these weaknesses — before they reach production.

Code must be maintainable without AI

What happens when the LLM model changes or the AI service goes down? My code follows Clean Architecture principles and is human-readable — today and in 5 years.