I put machine learning into real operations.
I'm at my best finding a working solution when no one yet knows how to tackle the problem. I've shipped production ML on a Fortune-500 sales floor at Procter & Gamble, led AI projects at GF Machining Solutions, and built — and still run — a CHF 500K+ e-commerce where restocking, support, pricing and anomaly detection largely run themselves.
Before industry, I did research at IDSIA, where I published. I hold an MSc in Artificial Intelligence (IDSIA). Based in Lugano; open to AI, ML and automation roles around Zurich, Zug and Basel, and remote.
The same energy, pointed at people.
Outside of work I've spent years in the community: a first responder with the Samaritans, an educator in scouting, and part of the Global Shapers Community (World Economic Forum). I gravitate to projects that put technology and effort where they actually help people.
That instinct shows up in my work too — social-sport programs that give kids a safe place to move, and data journalism that holds local media to account. Same drive as the engineering; different scoreboard.
Things I've shipped.
A few highlights where ML or automation made a measurable difference. Every number is real and defensible.
Real-time recommender for Amazon sales
A real-time ML alert/recommender that surfaces anomalies and opportunities in Amazon sales data and suggests the next action.
ML lead-scoring that won the room
An ML lead-scoring tool that beat the incumbent and earned executive sponsorship for further AI work.
Automating the finance work nobody wanted
An ETL + web-scraping automation that reproduced a manual finance task end to end, saving thousands of hours a year.
An e-commerce that runs itself
Built a coffee e-commerce from zero, then automated its operations end to end — restocking, support, pricing and anomaly detection.
What I do when no one's paying me.
A through-line across my life: show up, take responsibility, and put effort where it helps.
Samaritans — first responder
Trained first-aid volunteer, on call for community events and emergencies. Calm-under-pressure, the non-technical kind.
Scout educator
Years guiding kids and teenagers in scouting — responsibility, teamwork and the outdoors, one weekend at a time.
Global Shapers Community
Part of the WEF's young-leaders network, on local projects with a social and civic focus.
IdéeSport
Led social-sport programs giving kids a safe, free place to move and belong after school.
ex-President, GMSI
Led a community association — the unglamorous work of keeping people organised and motivated.
TEDxLugano · WWF · Impact Hub
Volunteer hands for events, environmental causes, and incubating impact-driven ventures in Ticino.
Where I've worked.
AI Sales Engineer — Napolicaffe.ch (FIRMY SA)
The AI/data engine of a live e-commerce P&L: automation platform for restocking, support, pricing and cost-anomaly detection.
Product Manager & Lead Developer — WeSherpa
Sole dev + PM; concept to working MVP, then a clean exit when funding didn't close.
eCommerce Manager — Napolicaffe.ch
Built the business from scratch to CHF 500K+; price-elasticity & demand-forecasting models; A/B testing (+70% engagement).
Data Scientist — Procter & Gamble
Real-time recommender across 5 brands/6 countries; automations saving ~4,000 hrs/yr; Product Owner leading a team of 5.
AI Solutions Lead — GF Machining Solutions
ML lead-scoring deployed in 3 countries under VP/CSO sponsorship; won the internal AI hackathon (NLP).
AI Research Assistant — IDSIA
Published research on AI-driven optimisation; built a RESTful CNN system and a 3D drone-trajectory tool.
Notes on shipping ML into the real world.
Essays in progress, drawn from things I've actually had to make work. (Drafts — titles are placeholders.)
Putting ML into a live P&L: what actually breaks
The gap between a notebook model and a system that quietly runs a business every day.
Automating the work nobody wants to touch
How to earn trust before you replace a manual process — and the ~4,000-hour lesson behind it.
Detecting media bias with NLP
Notes from "Sotto la Lente": analysing local news to surface slant in coverage.
Let's talk.
Open to AI, ML and automation roles around Zurich, Zug and Basel — and remote. The fastest way to reach me is email.