Hi there I'm Luca 👋
Hi 👋, I am Luca, a postdoctoral researcher at Fondazione Bruno Kessler (FBK) in Trento 🇮🇹, where I work in the Machine Translation Unit. My current research focuses on Multimodal Large Language Models, with a special interest in Speech.
I completed my PhD in Computer Science at the University of Pisa 🇮🇹 in July 2025, with the highest grade (summa cum laude), under the supervision of Prof. Anna Monreale and Dr. Nicolas Kourtellis. During my PhD I worked on AI safety for Federated Learning: privacy, fairness and explainability. I have authored papers at venues such as FAccT and ECAI (see my research), spent a year as a research intern at Telefónica Research in Barcelona 🇪🇸, and collaborated with Johannes Gutenberg-Universität Mainz.
Outside the lab I am a speaker, a community builder (Superhero Valley and Pisa.dev) and a podcaster (310+ episodes).
My path #
- 2025 – Present
Postdoctoral Researcher
Fondazione Bruno Kessler (FBK), Machine Translation Unit · Trento, Italy 🇮🇹- Working on Multimodal Large Language Models, with a special interest in Speech.
- Feb 2025
Devroom Manager
"Low-Level AI Engineering & Hacking" Dev Room @ FOSDEM 2025 · Brussels, Belgium 🇧🇪- Co-organized a dedicated track for AI practitioners and researchers: call for proposals, speaker selection, schedule and logistics.
- Mar 2023 – Mar 2024
Research Intern
Telefónica R&D, System AI Lab · Barcelona, Spain 🇪🇸- Designed PUFFLE: a method to balance privacy, utility and fairness in Federated Learning (ECAI 2024, patent application submitted).
- Supervised by Nicolas Kourtellis, Mikko Heikkilä and David Solans. Partially funded by the SPATIAL EU project.
- Jul 2022 – Present
Co-founder and Core Organizer
Pisa.dev · Pisa, Italy 🇮🇹- A community for developers in Pisa. 20+ meetups and social events organized.
- Nov 2021 – Jul 2025
Ph.D. in Computer Science, summa cum laude
University of Pisa, KDD Lab · Pisa, Italy 🇮🇹- Thesis: "Beyond Model Accuracy: Building Trustworthy Federated Learning Systems", advised by Anna Monreale and Nicolas Kourtellis.
- Privacy, fairness and explainability in Federated Learning. Partially funded by the Horizon Europe project TANGO.
- Started collaborations with Telefónica Research and Johannes Gutenberg-Universität Mainz. Supervised 3 master's theses, one of which led to a paper at ECAI 2025.
- Jun 2021 – Oct 2021
Research Assistant
University of Pisa · Pisa, Italy 🇮🇹- Won a research scholarship to extend my master's thesis on semantic enrichment of XAI for healthcare, later published at Discovery Science 2023.
- Oct 2020 – Present
Co-founder and Community Manager
Superhero Valley · Pisa, Italy 🇮🇹- A community of 2000+ computer scientists that bridges university and industry. 100+ members have secured roles in Big Tech.
- Sep 2020 – Apr 2022
Teaching Assistant
University of Pisa · Pisa, Italy 🇮🇹- Programming and Algorithmics, "Laboratorio 1" (JavaScript and TypeScript), and freshmen tutoring.
- Mar 2020 – Jan 2021
Flutter Developer
Tocket · Remote 🇮🇹- Built the Tocket app for buying tickets to online and in-person events, plus internal APIs and dashboard. 20k+ downloads on the App Store and Google Play.
- Oct 2019 – Mar 2020
Erasmus+ exchange
Technical University of Munich · Munich, Germany 🇩🇪- Courses in Machine Learning, Imaging and Visualization, and Cloud Databases.
- May 2019 – Present
Co-creator and Podcaster
PointerPodcast · Pisa, Italy 🇮🇹- A tech podcast with 280+ episodes and 70+ guests from academia and industry.
- 2018 – 2021
M.Sc. in Computer Science, 110/110 cum laude
University of Pisa · Pisa, Italy 🇮🇹- Thesis: "Semantic enrichment of XAI explanations for healthcare", supervised by Anna Monreale and Cecilia Panigutti.
- 2014 – 2018
B.Sc. in Computer Science
University of Pisa · Pisa, Italy 🇮🇹- Thesis: "Construction and Analysis of the Ethereum Transaction Graphs", supervised by Laura Ricci and Damiano Di Francesco Maesa.
Software Projects #
- PUFFLE: a regularization-based method that balances privacy, utility and fairness in Federated Learning, with custom fairness requirements defined by practitioners (Mar 2023 – ongoing).
- Fire360: a fast local explainer for tabular data that uses a single synthetic neighborhood instead of costly per-instance generation (Feb 2025 – Mar 2025).
- FastShap++: an extension of FastSHAP to Federated Learning with Differential Privacy to protect client data during explanation (Oct 2024 – Feb 2025).
Podcast and Communities #
🎙 I have always been attracted to communities.
Between 2013 and 2015, I worked as an editor for Biteyourapple.net, writing more than 1,000 blog-posts, recording videos and participating in the Bitecast podcast. I am one of the PointerPodcast co-founders. PointerPodcast is a podcast about technology and innovation, published since 2019. In some episodes, we interview software engineers and researchers from academia and industry. I have also co-founded two communities. The first is Superhero Valley, which fills the gap between university and industry by making students aware of the opportunities offered by big tech companies. The second is Pisa.dev, a local community that organizes meetups for developers and computer scientists in Pisa 🇮🇹.
🍦 In my free time I enjoy acting as a critic of TV shows, mainly in my podcast.
Volunteering #
GDG Pisa Devfest 2026: I was one of the helpers of GDG Pisa Devfest 2026.
GDG Pisa Devfest 2025: I was one of the helpers of GDG Pisa Devfest 2025.
Co-organized the “Low-Level AI Engineering & Hacking” Dev Room at Fosdem 2025, a dedicated track for AI practitioners and researchers. Curated the schedule, coordinated speaker selection, and managed the call for proposals to ensure a diverse and high-quality lineup of talks.
GDG Pisa Devfest 2024: I was one of the helpers of GDG Pisa Devfest 2024.
GDG Pisa Devfest 2023: I was one of the helpers of GDG Pisa Devfest 2023. I was one of the stage managers and helped record interviews with attendees and speakers. The video of the interviews is here.