I grew up close to the mountains in northern Italy, which is probably where both halves of this come from: a
liking for systems with hard constraints, and a complete lack of interest in answers that cannot be checked.
I studied industrial engineering in Trento, spent an Erasmus semester at TU Wien learning to work in German,
and then did a master's in mechanical engineering at Politecnico di Milano, specialising in materials and
manufacturing. My thesis was a load monitoring system for aeronautical structures — an inverse problem,
solved with a calibration matrix, verified numerically and experimentally. It later became a journal
publication. Looking back, it set the pattern for everything I have done since: sparse noisy sensors, a
physical model doing the heavy lifting, and statistics handling what the model cannot.
After graduating I became a resident engineer for Industrie Saleri Italo, embedded at the
BMW Motoren plant in Steyr. Four years designing and industrialising thermal management
systems and water pumps, from first concept to series production — and, just as importantly, four years of
standing in a plant, supporting a customer in German and English, and finding out how products actually fail.
You learn a lot about data quality by being the person who has to explain a defect.
Somewhere in there it became obvious that the questions I found most interesting were statistical ones. So I
enrolled at JKU Linz for a master's in artificial intelligence and did it while working full
time — mathematics, statistics, machine learning, deep learning, generative models, and a stint tutoring the
"Hands on AI" course. Average grade Sehr gut, and more usefully, the theory to back up what I had
been doing by intuition.
Since June 2023 I have been the lead data scientist in the central manufacturing department at
thyssenkrupp Automotive in Liechtenstein. Computer vision on assembly lines, anomaly detection at
the scale of millions of datapoints per day, an on-premise RAG assistant over the company wiki, hybrid
physics/ML models that cut scrap, Bayesian networks for uncertainty quantification. End to end, every time:
acquisition, storage, labelling, model, validation, deployment, monitoring. I have also helped shape the
company's digital transformation and AI strategy, collaborated with FH Vorarlberg and OST, supervised a
master's thesis as industrial tutor, and represented the company at Bodenseegespräche 2024 and ICML 2024 in
Vienna.
I am now looking for the next place to do this kind of work.
