Nice to meet you!
I'm Matteo.

Project Mobility Student @ ETH Zurich.

Pretraining foundation models for surface electromyography (sEMG) biosignals using transformer-based architectures. Explored Vision Transformers (ViT), including Channel-ViT, ViT-MAE variants, and SimMIM-style pre-training. Investigated efficient attention mechanisms and mixed-precision training to maximize computational efficiency and scalability. Conducted extensive downstream evaluations across several key tasks: gesture classification, finger movement estimation, silent speech recognition, and voiced silent speech.

Joined Outlier AI as an AI Training for Italian Writers.

Contributed to the Genesis Multimodal project, focusing on the safe and effective alignment of Multimodal Large Language Models (MLLMs) through Reinforcement Learning from Human Feedback (RLHF). Responsibilities include designing evaluation protocols, assessing model outputs for ethical compliance, safety, and fairness, collaborating with teams to refine alignment strategies, and analyzing model behaviors to ensure adherence to responsible AI principles. Employed highly specific prompts with several constraints to intentionally elicit failure responses from state-of-the-art models and then proceeded to identify model weaknesses by manually rewriting and refining model outputs to meet all task requirements.

Research Fellowship in Embedded Systems for Assistive Technologies @ University of L'Aquila.

Research activities in the field of Embedded Systems for Assistive Technologies. Engaged in hands-on research projects focusing on embedded systems with specific focus on Versal VCK190 by AMD/Xilinx. Contributed to the design and development of innovative solutions aimed at enhancing accessibility and improving the quality of life for users with special needs. Acquired significant skills in implementing embedded systems and collaborated effectively in an advanced research environment.

MSc in Artificial Intelligence @ University of Bologna.

Currently enrolled in the MSc in Artificial Intelligence program at the University of Bologna. Engaged in advanced coursework in Artificial Intelligence, Machine Learning, and Deep Learning. Acquiring in-depth knowledge of AI algorithms, techniques, and methodologies. Developing hands-on skills in AI programming, data analysis, and machine learning model development. Engaging in research projects and collaborating with peers in an advanced academic environment.

BSc in Computer Science with focus on Data Science @ LUMSA University.

Thesis focused on evaluation and modeling of timing performance of Neural Networks running on Versal AI Engine in presence of contention on shared resources.