I’ve been taking things apart since I was a kid: computers, gadgets, anything with screws I wasn’t supposed to remove. Most of the time I even managed to put them back together.

That curiosity never went away, it just found bigger things to take apart. I spent more than a decade in data engineering, making SQL Server and the cloud platforms around it faster, designing systems that don’t collapse at scale, and occasionally explaining why “just put it in Excel” isn’t always the answer.

Today I lead teams that bring generative AI to production: retrieval-augmented generation, AI agents and LLM applications that have to survive real users and real data. A big part of the job is translating what the business needs into technical requirements an engineer can build against. The data engineering never stopped mattering. Most GenAI problems are data problems in a new hat: what you retrieve, how fast, and how much memory it costs.

I still build, because I learn by doing and then share what I find. That’s how raggio happened, a small RAG vector database for machines that can’t spare gigabytes of RAM. Before it came the SQL Server Diagnostic Book for Azure Data Studio and a Power BI report for SQL Server Agent jobs.

I speak at conferences, and I organize them too: the Ticino Data Conference and the Swiss chapter of AI Saloon. On YouTube I post webinars, tech interviews and other things from my work.