Federico in a nutshell:
A data science journey in healthcare
If you’re reading this, chances are we already share something: curiosity. I like learning new things, and even more when they can make healthcare a little better. That’s what I do as a data science researcher in healthcare, but the part I enjoy most is the exchange, sitting with people from very different fields, understanding how they see a problem, and letting a good conversation turn into a better question. You’ll often find me at workshops and seminars doing exactly that. And I care a lot about how we work together: I believe an honest, generous environment is what lets people do their best work.
Places and teams where I’ve worked on a shared problem. The framing here is the challenge, not the title.
Keeping sensitive data in-house with local LLMs
Medicilio · Technology advisor · 2025 · Milan
Cutting administrative workload without handing sensitive data to outside providers, through internal documentation workflows built on local LLMs so the data stays in-house.
Making AI’s energy cost measurable
Broken Pot · Data consultant · 2025 to 2026 · Padua
Turning energy efficiency from a broad ambition into measurable cuts in consumption. My part sits on the data side.
Data and machine learning for robotic surgery
Needleye Robotics · External consultant · 2024 to present
Supporting an AI robotic-surgery startup on the data and machine-learning side of a hard clinical problem.
Turning everyday hospital data into evidence
Padua University Hospital · Healthcare data analyst · current
Alongside clinicians in the medicine department, making the data a hospital already produces genuinely usable for decisions about care.
Cross-border research
Biostatistics and surgery · 2021 to 2025
Biostatistics with Mainz University (Germany, with Bari University Hospital) and surgery with Ohio State / Wexner Medical Center (USA).
List of published papers
Multimodal models when data is scarce, sensitive and messy
A popular myth says useful machine learning needs big, clean datasets. Healthcare rarely gives you either, since data is fragmented, expensive to label, and locked down by privacy rules. My thesis took on exactly that problem, how to get models that still perform when the data is limited and arrives in very different forms.
In practice that meant working with multimodal data, structured tables, unstructured text and images together, and using transfer learning to carry knowledge across domains when labelled data is scarce. It also meant combining classical statistics, including survival analysis, with deep learning, and knowing when each one wins, all while treating pre-processing as seriously as modelling, because in low-data settings one weak step upstream quietly ruins everything downstream.
The takeaway is simple. Even with little data, the right approach turns fragmented healthcare data into decisions clinicians can act on. Happy to share a copy, just get in touch.
├─💼 Job Experience
│ ├─ Padua University Hospital · Healthcare data analyst, Medicine department
│ ├─ Broken Pot · Data consultant, energy efficiency
│ ├─ Medicilio · Technology advisor, local LLMs for internal documentation
│ ├─ Needleye Robotics · External consultant for an AI robotic surgery startup
│ ├─ Mainz University (DE) · Healthcare analytics consultant
│ ├─ ThyssenKrupp Component Technologies division · Intern, sales department
│ ├─ University of Turin · Intern, Data Science
│
│ └─📚 Education
│ ├─ University of Verona - MSc Data Science
│ └─ University of Turin - BSc Statistics & Economics
│
│ ├─📦 Projects
│ ├─ Cuochi Senza Fuochi - Founder of a group with member coming from different city or country. The format aim is to meet, know each other and share experience while cooking dessert without having a kitchen or experience in Padua
│
└─💡 Volunteering
│├─ Medics with Africa CUAMM - Broadcaster and host
│├─ Medium - Blogging about Data Science
├─ Centro Studi Italia Canada - Writer about Pan-Canadian AI initiative