Data, AI and transformation
under executive responsibility.
More than 16 years across strategy, architecture and operations — from building reliable data platforms to leading data and AI functions at C-level.
Data & AI leadership
Built and scaled data and AI functions from the ground up
Operationalized AI strategies — from target picture to production
Established AI governance under the EU AI Act in live operations
Held C-level responsibility for data organizations
Organizational design for data & AI
Built and structured teams
Microteaming framework for lean, effective units
Defined roles and decision rights
From roadmap to execution
Diagnosed and developed data culture
Translated AI roadmaps into operational measures
Integrated processes and guided change
What I have built with
AI architecture design
End-to-end design: data pipeline, model hosting, application layer. Build vs. buy, cloud vs. on-premise, open source vs. proprietary — decided together with IT, business units and executive management.
From practice: end-to-end architecture for fully automated document processing — from intake to closed case.
Retrieval-Augmented Generation (RAG)
LLM-based knowledge systems on company data — contracts, technical documentation, regulatory requirements. With source references, currency and traceable answers.
From practice: RAG system for a compliance department — answers on internal policies within seconds, with sources.
AI-driven data enrichment
LLM-based enrichment of incomplete master data — classification, categorization, filling missing fields. Half-complete records become usable ones.
From practice: enrichment of 200,000 existing records — automated classification along business criteria, with sample-based validation.
AI-driven data quality
Anomaly detection, duplicate matching and plausibility checks with LLMs and ML — on master and transactional data. Data quality becomes routine instead of a one-off campaign.
From practice: AI-assisted duplicate detection in the CRM — manual review becomes suggested merges with reasoning.
Churn prediction (classical ML)
Gradient boosting and logistic regression instead of a black box. Feature engineering the business understands — and sales actually uses. Weekly routine rather than AI hype.
From practice: churn model on 15,000 existing customers — weekly risk lists for account managers.
Engagement scoring & lead prioritization
Activity data — logins, feature usage, support tickets — condensed into an operational score. Clear thresholds trigger retention, cross-sell or rescue measures. Updated daily, directly in the CRM.
From practice: engagement score coupled to Salesforce — integrated into customer success workflows.
How I frame these topics
If you want to read regularly about the topics on this page: in "Techne & Polis" I take stock every Thursday of what is moving in technology and what follows from it for work, the public sphere and institutions. Free, unsubscribe anytime — published in German.
For talks and inquiries: Contact