ABOUT ME
I have worked at the intersection of data science, information technology, and research for over 25 years. I have a background in computer science and earned a doctorate in innovation and strategy from the University of Innsbruck. My work combines research with practical application. I regularly publish in both academic journals (1) and practice-oriented media (2, 3, 4, 5).
My research interests are quantitative methods in market research: statistical models that find patterns across groups in large data sets. Over the years this led to an AI-supported analysis platform with interactive dashboards for large-scale studies.
On the technology side I am fascinated by domain-specific languages (DSLs) and the question how far a domain model can carry an application. This is the work I do with my partner at die modellwerkstatt, and more recently it includes putting large language models (LLMs) to work inside those domain models.
I live with my family in Vorarlberg, Austria.
WHAT KEEPS ME BUSY
Research and data
Choice-based conjoint, Kano analysis and multivariate methods are the tools I reach for when the question is what people actually prefer. I have designed many online surveys and learned that clear structure and good sampling matter more than clever statistics. I care about visualizations that let people without a statistics background see the result.
Languages, generators and agents
At die modellwerkstatt we build werkbank, a Java toolchain on JetBrains MPS, and MoAI, which lets coding agents work directly in those models. My part is language design, runtimes and the integration of LLMs through the Model Context Protocol (MCP), agent skills and validated workflows. Before that I built web and enterprise applications with Django, Spring and Vaadin, and I still enjoy a good ORM discussion.
WHERE TO FIND ME
The Rhine Valley is one of the most beautiful places for me.
If you want to get in touch, write to daniel.stieger@modellwerkstatt.org. For work with die modellwerkstatt, the company site is the better starting point.