Information
📚 Organization | Politecnico di Torino |
📆 Period @ ESA Φ-lab | September 2025 – December 2025 |
🌍 Project @ ESA CIN | ⚠️ Project is ongoing - L0-L1 PhiSat-2 Processing with Deep Learning |
📍 GitHub | https://github.com/ThePiromaximus |
🔨 Linkedin | https://www.linkedin.com/in/gabriele-inzerillo-6411b5147/ |
📝 Publications | Google Scholar |
Bio
Gabriele Inzerillo is a PhD student in Computer Vision and Deep Learning. His professional path began at the Istituto Tecnico Industriale, where he spent five years developing a solid foundation in programming, IT systems, and databases. He went on to earn a BSc in Computer Engineering at the University of Palermo, before moving to Turin to complete an MSc in Computer Engineering at Politecnico di Torino, where he strengthened his skills in software engineering and artificial intelligence.
For his MSc thesis, carried out at Serco Group, he created two remote-sensing image datasets from the Proba-V and Sentinel-2 satellites and adapted a state-of-the-art super-resolution model to enable effective cross-sensor image enhancement. After graduating, he worked for a year as an applied scientist, focusing on computer vision models for remote sensing imagery and on the development of efficient models through quantisation techniques and deployment on low-power hardware.
Current Role
Gabriele is currently pursuing a PhD in Deep Learning through a joint project between Politecnico di Torino and the European Space Agency within the OSIP framework. His work focuses on designing efficient deep learning models capable of running onboard satellites to perform vision tasks with low latency, balancing research challenges with practical technical and hardware constraints—positioning his work between theoretical and applied research.
Areas of Expertise
He specialises in Deep Learning and Computer Vision, with a focus on efficiency for edge AI and onboard processing of remote-sensing data. His academic and professional experience has strengthened both his theoretical and practical skills in self-supervised learning, learned compression, change detection, and image alignment/coregistration, all applied to satellite imagery. These capabilities support his work on PhiSat-2 image translation using deep learning models to enable faster, more efficient retrieval of L1 data from L0 imagery, and give him a strong understanding of the specific challenges inherent to remote-sensing applications.
Vision for the Future
Gabriele aims to advance AI for both satellite and natural images, transforming raw pixels into timely and trustworthy insights. He envisions a future in which ultra-efficient models run on-device or on-orbit for immediate responses when connectivity is limited, complemented by more powerful models operating on the cloud for deeper analysis. His goal is to create a seamless interaction between these two layers, enabling fast decisions when necessary and comprehensive ones when possible, ultimately making Earth observation and computer vision more actionable, sustainable, and widely accessible.