Expertise & method
Compliance that scales instead of slowing you down
I apply proven principles from software engineering to AI governance. The result: compliance becomes testable, verifiable and automatable – instead of disappearing into documents.
The method
Specification-Driven Compliance
Three principles decide whether compliance stays a cost driver – or becomes an enabler.
Testability over interpretation
Rule-based decision structures instead of vague policies. What is testable is decidable – and can be automated.
Systems over documents
Automated approvals instead of manual case-by-case review. Compliance runs inside the system instead of gathering dust in folders.
Evidence over "trust me"
Every decision traceable and auditable. No blind flying, no AI hallucinations – solid evidence at the push of a button.
Why this makes the difference
Two ways to run AI compliance
One adds manual work with every AI system. The other keeps paying off once the setup is done.
- Policies get written and outputs are reviewed by hand
- Every new case means new manual work
- Effort grows linearly with the number of AI systems
- Evidence is laborious and full of gaps
- Rules are defined as a testable specification
- The system checks and documents itself
- Effort is spent once at setup, then it scales
- Every decision is auditable automatically
TUMAKI works consistently towards class B – and gets you there without stalling your day-to-day operations.
Qualifications
What the method is built on
Certifications & roles
- TÜV Rheinland Certified AI Consultant
- Certified Data Protection Officer
- Compliance Expert, Koerting Institute
- Lean Six Sigma Master Black Belt
- Speaker at industry conferences
Regulatory focus
- EU AI Act (risk classes, Art. 9–15)
- GDPR for ML/AI (DPIA, training data)
- MaRisk & BAIT / VAIT for the financial sector
- BaFin / EBA requirements
- NIS-2, EU Data Act, Digital Omnibus
Interested?
Let's look at your AI landscape
In an intro call I show how Specification-Driven Compliance applies concretely to your systems.