PoliToIndustrial Production and Technological Innovation Engineering
Politecnico di Torino · Turin, Italy
The foundations: systems, processes, and engineering rigor.
I like figuring out how things work, then finding ways to make them better.
EXPLOREI'm an engineer, researcher, innovator, and co-founder, but I've never been interested in staying inside only one of those boxes.
I like understanding how things work, asking too many questions, and turning ideas into things that actually exist. These days, most of those questions happen to be about AI.
Five degrees, four countries, and a lot of curiosity along the way.
PoliToPolitecnico di Torino · Turin, Italy
The foundations: systems, processes, and engineering rigor.
UICUniversitat Internacional de Catalunya · Barcelona, Spain
Technology seen from the business side: strategy, organisation, and people.
KTHKTH Royal Institute of Technology · Stockholm, Sweden
The first year of the master's at Sweden's largest technical university, ranked in the world's top 100: robotics, control, and systems that act in the real world.
EURECOMEURECOM · Sophia Antipolis, France
The second year of the master's at EURECOM, a leading French graduate school and research centre in digital sciences, in Sophia Antipolis: France's Silicon Valley. Here I focused on artificial intelligence.
EITEuropean Institute of Technology · Stockholm · Sophia Antipolis
A programme co-founded by the European Union across top European technology universities, where innovation and entrepreneurship travel together.

Building agentic systems that bring context, knowledge, and action into business operations. The rest, for now, intentionally stays high level.

Evaluating, optimising, and efficiently deploying LLMs. This is where WISEBench started: a tool for picking the right model across quality, latency, cost, and energy use.

AI agents that automate manual workflows: design with LangChain, fine-tuning across model sizes, and research into what agents can actually do.

Automating accounting and reporting processes: started as an internship, ended up leading the project. Credit risk analysis and data-driven decisions.
Things built and questions made public, so they can become conversations.
01Project
Thesis · framework
A framework for choosing the right model across quality, latency, cost, compute, and energy.
↗02Paper
EvalLLM 2025
Controlled errors to understand which metrics actually detect factual degradation.
↗03Project
Microsoft · Italgas · 2022
Built by a multidisciplinary team and awarded at the ChangeBOT challenge.
↗04Article
Medium · NeuroAI
A simple mental model for separating useful analogies, metaphors, and hype.
↗05Meta-analysis
Medical AI
A critical view of AI diagnostic reliability in mammography, compared with human radiologists.
↗→
Projects, papers, and writing in one full list.
See all ↗Tools, methods, and territories I spend time in.
Research, AI products, new ideas, or simply an interesting conversation.