Understand technology. Move business forward.

My further training as an AI Implementation Architect, completed in 2026, adds current AI expertise to many years of project and business application experience. It covers use-case assessment, Azure AI and Azure OpenAI, APIs and automation, as well as governance, data protection and regulatory frameworks.

The order matters to me: understand the problem and the value first, then choose the right technology. Sometimes AI is the answer; sometimes it is a well-designed API, conventional automation or simply a better workflow.

In 2005, around 350 dealer locations submitted trial balances on floppy disks by post. The data centre loaded and processed the disks manually and faxed results back. Unreadable disks and repeated corrections made the mandatory benchmarking process inflexible. Around 650 rules checked plausibility; the assessment also mattered for new vehicle margins.

I developed the portal vision and design, supported coordination and introduction, and trained all dealers. Uploads were processed in around 15 minutes. Dealers could correct data independently; the manufacturer gained access to current results, submission rates, dealer histories and cross-dealer analyses. Telephone hotline workload fell significantly.

Open project story →

In the Toyota DMS environment, I worked in first- and second-level support on financial accounting, vehicle operations, inventory and dealer-related ERP and CRM topics. A particularly formative experience was helping numerous dealers move from a contribution-margin approach to cost accounting. I supported around 100 dealer migrations in this project.

After each migration, the data had to be checked from an accounting perspective. If the balance sheet did not reconcile, looking at an error message was not enough: trial balances, accounts, account classes, posting logic and the discrepancy itself had to be traced systematically. Some cases took several days and multiple rounds of coordination with the dealer's accounting team.

An Excel-based tool using pivot tables and structured analyses was developed during the project to help narrow down discrepancies faster. A case was only closed when the balance sheet reconciled — not when the ticket had merely been passed on.

In 2002, I supported around 600 dealerships together with a ten-person team. DRACAR+ and OPTIMA21 ran on Windows, AD2000 on Unix. The industry ERP systems had around 10,000 parameters; Crystal Reports was used for embedded forms.

I worked with German support, the Product Owner in Germany and the development team in Denmark, formulated requirements, reviewed specifications and supported pilots. For the data service supplying around 2,000 dealers, I designed online distribution of parts price master files and a paid managed service for automatic updates with email confirmation. Based on my assessment at the time, subsequent implementation reduced ongoing effort from around two full-time positions to 0.2 FTE. Other work included a web-based DMS concept and the ADP Dealer Services website.

Open project story →

While self-employed, I ran six online shops with marketplace connections to Amazon, eBay and Yatego. Around 1,000 products, up to 40 orders a day and monthly revenues usually between €8,000 and €10,000 — reaching around €12,000 in good months — quickly turned technical decisions into commercial ones.

JTL-Wawi was the central inventory management system. Marketplace orders were imported automatically, stock levels synchronised across channels and documents generated. At the same time, our own stock and dropshipping had to work together: supplier data was updated regularly, while other product ranges were stored, packed and shipped in-house.

When standard features were not enough, additional modules were integrated selectively. My understanding of inventory management, interfaces and process automation therefore comes not only from projects, but also from personal responsibility for stock, margins, shipping, customer service and profitability.

The starting point was paper timesheets for around 700 employees and a process chain linking workforce scheduling, assignment planning, employee communications, transport, time recording and accounting handover. I documented the existing end-to-end processes, derived business requirements and contributed to the functional specification and selection of what became the zvoove solution.

I reviewed the preconfigured roles and workflows from a business perspective and manually tested all intended user roles. Implementation was deliberately phased: first with around ten employees, then with about 50. Only after these practical tests did the solution move into a broader rollout.

The solution ultimately went live for around 700 employees at the Dresden, Leipzig and Erfurt branches included at the time. Paper timesheets were fully replaced, and assignment and time data could be processed digitally through to accounting handover.

As an on-site support specialist in the ASOS fulfilment environment, I supported and monitored around 700 printing and copying systems. In logistics, availability is not an abstract metric: a faulty or misconfigured device can immediately hold up a real workflow.

Incidents therefore had to be prioritised by operational impact, root causes identified quickly and solutions implemented with minimal disruption to ongoing fulfilment operations. The work sat directly between users, technology and day-to-day operations.

This experience sharpened my focus on operational stability and practical troubleshooting. The most technically interesting solution is not always the best in a live environment. What matters is keeping the process running.