The scientific publication “eva: Continuous Learning for STEM Students”, developed by three Master’s students at Politecnico di Milano under the supervision of Prof. Marco Santambrogio (Department of Electronics, Information and Bioengineering) and PhD candidate Laura Ginestretti (Information Technology), has been selected for the international IEEE EDUCON 2026 conference, held in Cairo, where it was awarded first prize in the Student Poster Competition.
The project is part of the LEONARDO initiative at Politecnico di Milano and was originated from the final challenge of the course “Creativity, Science and Innovation”, titled “The Creative Science Behind Student Life: How Can Language Models Improve Your Day?”.
During the final event held on June 25, 2025, the team presented an interactive Proof of Concept, which was evaluated through a pitch and live demonstration by a panel of engineers and researchers. The project subsequently received further recognition by ranking among the top 100 initiatives in the Switch2Product entrepreneurial acceleration program at Politecnico di Milano.
The team consists of three Master’s students:
- Matteo Minotti, with a background in Biomedical Engineering and currently enrolled in Nuclear Engineering, contributed to research and data analysis, with a focus on advanced computational methods for particle accelerator physics;
- Riccardo Masetti, a Computer Science and Engineering student and future intern at the European Space Agency (ESA), developed the platform’s core functionalities using technologies such as PostgreSQL, React, and Django;
- Pietro Carrucciu, also a Computer Science and Engineering student, was responsible for the design of the user experience and user interface.
From a scientific perspective, the paper addresses the phenomenon of cramming, namely the tendency among STEM students to concentrate studying immediately before exams. To mitigate the limited presence of continuous assessment during the semester, the “eva” platform integrates principles of Self-Regulated Learning with the use of large language models. The system enables the automatic generation of continuous, multimodal assessment sessions based on user-uploaded materials, promoting distributed practice, increased awareness of individual progress, and improved long-term retention.