Loading...

Hi, I'm Daniele Falcetta

Ciao, sono Daniele Falcetta

AI Research Scientist | 3D Computer Vision & Data-Efficiency

AI Research Scientist | 3D Computer Vision & Data-Efficiency

I specialize in Data-Centric AI, architecting Active Learning frameworks and scalable infrastructure to solve the 'Data Scarcity' bottleneck in high-dimensional 3D vision. My research bridges the gap between deep learning theory and production engineering, leveraging complex topological data (medical imaging) to build robust, label-efficient models that significantly reduce annotation costs.

Specializzato in Data-Centric AI, progetto framework di Active Learning e infrastrutture scalabili per risolvere il problema della 'Data Scarcity' nella visione 3D ad alta dimensione. La mia ricerca colma il divario tra la teoria del Deep Learning e l'ingegneria di produzione, utilizzando dati topologici complessi (come l'imaging medico) per sviluppare modelli robusti e label-efficient che riducono drasticamente i costi di annotazione.

Daniele Falcetta

About Me

I am an AI Research Scientist (PhD Candidate) specializing in Data-Efficient Deep Learning and 3D Computer Vision. My focus is solving the 'Data Scarcity' bottleneck by architecting scalable Active Learning frameworks and MLOps infrastructure that drastically reduce labeling costs while maximizing model performance.

My research on Data-Centric AI for complex 3D topologies has achieved State-of-the-Art results, securing two 9% acceptances at MICCAI 2025. I bridge the gap between theoretical research and production engineering, using high-dimensional medical imaging as a high-complexity sandbox to solve universal computer vision problems.

With a Double Degree MSc. in Data Science and a background in Biomedical Engineering, I am passionate about deploying robust, safety-critical AI systems in real-world environments.

Sono un ricercatore in IA (dottorando) specializzato in Data-Efficient Deep Learning e 3D Computer Vision. Il mio obiettivo è risolvere il problema della 'Data Scarcity' progettando framework scalabili di Active Learning e infrastrutture MLOps che riducono drasticamente i costi di annotazione massimizzando al contempo le prestazioni dei modelli.

La mia ricerca sulla Data-Centric AI per topologie 3D complesse ha raggiunto risultati allo Stato dell'Arte (SOTA), ottenendo due accettazioni Top 9% alla conferenza MICCAI 2025. Colmo il divario tra ricerca teorica e ingegneria di produzione, utilizzando l'imaging medico ad alta dimensione come una 'sandbox' ad alta complessità per risolvere problemi universali di Computer Vision.

Con una doppia laurea magistrale in Data Science e un background in Ingegneria Biomedica, mi appassiona lo sviluppo e il deployment di sistemi AI robusti e safety-critical in ambienti reali.

Skills & Expertise

Deep Learning 3D Computer Vision Geometric Deep Learning Active Learning MLOps & Infrastructure PyTorch Python Domain Adaptation Collaborative AI

Tech Stack & Toolbox Stack Tecnologico

ML & Deep Learning ML & Deep Learning
🐍 Python 🔥 PyTorch 🧠 TensorFlow 📊 Scikit-learn 🔢 NumPy 🐼 Pandas
3D Vision & Medical Imaging Visione 3D & Imaging Medico
🏥 MONAI 🔬 SimpleITK 👁️ OpenCV 🧠 Nibabel 📐 VTK
MLOps & Infrastructure MLOps & Infrastruttura
🐳 Docker 📦 Git 📈 WandB 🖥️ Slurm 🐧 Linux 💻 VS Code
Data & Collaboration Dati & Collaborazione
📓 Jupyter 📝 LaTeX 🗄️ SQL 🐙 GitHub 🔄 CI/CD

Education & Experience

2023 - Early 2026

🔬 AI Research Scientist (PhD Candidate)

Sorbonne Université & EURECOM

Leading research on Data-Efficient Deep Learning. Architected "VesselVerse," a collaborative annotation infrastructure reducing labeling costs for 3D segmentation. Achieved SOTA results (Top 9% MICCAI) in Active Learning and Domain Adaptation.

Deep Learning Computer Vision MLOps
2023 - Present

👨‍🏫 Teaching Assistant

EURECOM

Teaching Assistant for MALIS (Machine Learning and Intelligent Systems) course and labs. Supporting graduate students in machine learning theory and practical implementations.

Teaching Machine Learning Mentoring
2022-2023

💼 AI Research Engineer Intern

SAP Labs, France

Developed MultiPath2Vec, an attention-based model for security vulnerability detection in code commits. Applied NLP and deep learning techniques to software security.

NLP Security Industry
2021 - 2022

📚 MSc. Data Science (Double Degree - Year 2)

EURECOM

Second year of Double Degree program with Mobility Scholarship. Specialized in Medical Image Analysis, Deep Learning, and Computer Vision with focus on healthcare applications.

Data Science Medical Imaging Scholarship
2020 - 2021

📚 MSc. Data Science (Double Degree - Year 1)

Politecnico di Torino

First year of Double Degree program in Data Science. Focus on Machine Learning and Computer Vision fundamentals with strong theoretical and practical foundation.

Data Science Machine Learning Double Degree
2017 - 2020

🎓 BSc. Biomedical Engineering

Politecnico di Torino

Bachelor's degree with honors. Selected for Young Talents - Honors Program (Top 1% of students). Founded strong technical and theoretical background in engineering and medicine.

Biomedical Engineering Top 1%

Featured Projects

Click on any project card to view its architecture pipeline

Clicca su qualsiasi progetto per visualizzare la pipeline dell'architettura

ISBI 2026 ERC CARAVEL (ERC)
📊

CaravelMetrics

Advanced metrics and evaluation framework for vessel segmentation analysis. Comprehensive toolset for assessing vascular structure segmentation quality. Accepted at ISBI 2026.

Medical Imaging Evaluation Metrics Vessel Analysis Python
MICCAI 2025 ERC CARAVEL (ERC)
🧠

VesselVerse Website

A comprehensive dataset and collaborative framework for vessel annotation, enabling multi-institutional collaboration in brain vessel segmentation research. Early Acceptance - Best 9%.

Medical Imaging Dataset Collaboration 3D Segmentation
MICCAI 2025
🎯

One-Shot Active Learning

Novel approach for vessel segmentation using active learning techniques to dramatically reduce annotation requirements while maintaining high performance. Early Acceptance - Best 9%.

Active Learning Deep Learning Medical Imaging
MELBA
🔬

Multi-Domain Brain Vessel Segmentation

Feature disentanglement approach for robust brain vessel segmentation across multiple imaging domains, published in Machine Learning for Biomedical Imaging.

Domain Adaptation Feature Learning Brain Imaging
BMVC 2023
🔄

A2V Framework

Semi-supervised domain adaptation framework for brain vessel segmentation via two-phase training angiography-to-venography translation. Presented at BMVC 2023.

Domain Adaptation Semi-Supervised Medical Imaging
Challenge
🏆

TopCoW Challenge

Benchmarking topology-aware anatomical segmentation of the Circle of Willis for CTA and MRA imaging modalities.

Benchmark Topology Vascular Imaging
SAP Labs
🔐

MultiPath2Vec

Attention-based model for security vulnerability detection in code commits, developed during research internship at SAP Labs.

Security NLP Code Analysis

Selected Publications

An Automated Framework for Large-Scale Graph-Based Cerebrovascular Analysis

⭐ First Author 📄 Conference ERC CARAVEL (ERC)

D. Falcetta, et al.

ISBI 2026, IEEE International Symposium on Biomedical Imaging

One-shot Active Learning for Vessel Segmentation

⭐ First Author 📄 Conference

D. Falcetta, et al.

MICCAI 2025, 28th International Conference on Medical Image Computing and Computer Assisted Intervention (Early Acceptance - Best 9%)

VesselVerse: A Dataset and Collaborative Framework for Vessel Annotation

⭐ First Author 🎤 Spotlight 📄 Conference ERC CARAVEL (ERC)

D. Falcetta, et al.

MICCAI 2025, 28th International Conference on Medical Image Computing and Computer Assisted Intervention (Early Acceptance - Best 9%)

Divergence-Aware training with automatic subgroup-mitigation for breast tumor segmentation

📚 Workshop 🏆 Best Paper Award

E. Poeta, L. Vargas, D. Falcetta, V. Marciano, et al.

MICCAI 2nd Deep Breast Workshop on AI and Imaging for Diagnostic and Treatment Challenges in Breast Care, 2025

Initial Analysis of Intracranial MR Angiography from the TwinsUK Large Cohort Twin Imaging Study

📝 Abstract 📄 Conference

J.O. Cleary, L.S. Canas, M.A. Zuluaga, D. Falcetta, et al.

Annual British Society of Neuroradiologists' Meeting (BSNR), 2025

Multi-Domain Brain Vessel Segmentation Through Feature Disentanglement

⭐ First Author 📰 Journal

Galati, F.*, Falcetta, D.*, Cortese, R., Prados, F., Burgos, N., & Zuluaga, M. A.

Machine Learning for Biomedical Imaging, 2025, Volume 3, Pages 477–495, ISSN 2766-905X

Benchmarking the CoW with the TopCoW challenge: Topology-aware anatomical segmentation of the Circle of Willis for CTA and MRA

🏆 Challenge

K. Yang, F. Musio, Y. Ma, ..., D. Falcetta, et al.

ArXiv preprint, 2024: arXiv-2312

A2V: A semi-supervised domain adaptation framework for brain vessel segmentation via two-phase training angiography-to-venography translation

📄 Conference

F. Galati, D. Falcetta, et al.

34th British Machine Vision Conference (BMVC), 2023

Presentations & Talks

Conference presentations, seminars, and research talks

Presentation Title

Open in Google Slides

Awards & Achievements

Recognition for academic excellence and research contributions

🎓
2025

Doctoral Consortium Finalist

MICCAI 2025 Doctoral Consortium

Selected among top PhD candidates worldwide
🎤
2025

Spotlight Presentation

MICCAI 2025 - VesselVerse

Top 9% Early Acceptance with Spotlight presentation
🏆
2025

Best Paper Award

MICCAI 2nd Deep Breast Workshop

Best paper in AI and Imaging for Breast Care
👥
2024-2025

Challenge Co-Organizer

TopBrain MICCAI 2025 Challenge

Leading international brain segmentation challenge
🌍
2024-2025

Research Dissemination

World AI Cannes Festival (WAICF)

Presenting AI research to global audience
✈️
2021

Mobility Scholarship Winner

18-months Double MSc. Degree @ EURECOM

Competitive international exchange program
📚
2020

Academic Scholarship Winner

MSc. @ Politecnico di Torino

Merit-based full tuition scholarship
2017

Young Talents - Honors Program

Top 1% students @ Politecnico di Torino

Exclusive program for exceptional students

Research Collaborations

Collaborating with leading research institutions across Europe

Get In Touch

I'm always open to discussing research collaborations, new opportunities in medical imaging and AI, or just chatting about interesting ideas in deep learning and computer vision.

Pipeline Overview

Project Pipeline