I came to engineering through the workflow, not the other way round.

The short version

I spent fifteen years as a product designer, mostly in places that were not glamorous: a corrugated cardboard plant in Valtellina, an iris-recognition hardware company in Bologna, a fashion CAD vendor whose users had been doing the same job since before the software existed. What those jobs teach you is that the documented process and the real process are different documents, and only one of them matters.

Then I co-founded Nimbo Jobs and ran it as CEO — an NLP and conversational system for matching temporary work. That is where the second half of the education happened: a model that works in a notebook and a model that survives a payroll are not the same artefact.

Now I build with LLMs directly. Agents, MCP servers, eval harnesses, native clients on the Claude API. I hold a master’s in AI, I lecture at Politecnico di Milano, and I speak at AI community events — including a keynote in Turin this year.

The book

The UX DeCode distils thirty years of interaction research — Nielsen, Tognazzini, the ISO ergonomics work, Gerhardt-Powals — into thirteen principles with observable pass/fail criteria. It is published in Italian.

It reads as a design book. It is really a diagnostic protocol: a repeatable way of finding where a workflow breaks before you have spent money automating the wrong part of it. That is the first two weeks of any deployment I have ever done. On Amazon

What I am looking for

A Forward Deployed Engineer role — or Applied AI Engineer, Solutions Engineer, Deployment Engineer; the titles vary and the job does not. Specifically: getting a system to work inside somebody else’s organisation, where the constraints are half technical and half political, and where being the person who does not yet know the domain is the starting position rather than a problem.

Based in Piacenza, Italy. Comfortable with travel. Italian and English.

I also take a small number of independent engagements. How that works