🔍 Vision Transformer for Anomaly Detection, 2025
At the ALCOR Lab (Sapienza University), under the supervision of Prof.ssa Irene Amerini, I worked on applying Vision Transformers (ViTs) to image anomaly detection. The project focused on leveraging transformer-based architectures for representation learning and evaluating their effectiveness against convolutional approaches, with experiments on benchmark datasets for industrial and security-related anomaly detection.

────────────────────────────────────────────────
♟️ HIVE playing engine, 2025
Together with my teammate Francesco Giordani, and under the supervision of Antonio Norelli, Matteo Frigo, and Matteo Ceccarello, I designed and implemented an intelligent agent capable of playing the Hive board game. The project involved experimenting with a range of adversarial search strategies, including Alpha-Beta Pruning, Lazy-SMP, Monte Carlo Tree Search (MCTS), and the integration of Large Language Models (LLMs), drawing inspiration from the AlphaGo family of systems.

────────────────────────────────────────────────
🌙 Tohoku Engineering Summer Program, 2025
During my visiting research in Prof. Kazuya Yoshida’s Space Robotics Lab at Tohoku University, I worked on an autonomous robot designed for lunar exploration. Together with Alessio Borgi, I developed a custom platform capable of navigating sandy, uneven terrain and interacting with target objects. The robot combined occupancy-map generation and Dijkstra-based path planning for navigation. For perception, we trained a vision system to detect objects and integrated visual servoing for fine alignment. Finally, a loader-style gripper enabled successful manipulation despite the challenges of loose, unstable surfaces.

────────────────────────────────────────────────
🦾 Franka Emika Research 3 – Projected & Reduced Gradient, 2025
Under the supervision of Prof. Alessandro De Luca, I carried out a project on redundancy resolution in robotic manipulators, developing an RViz simulation in ROS2 Humble for the Franka Emika Panda robot. The system implemented Projected Gradient (PG) and Reduced Gradient (RG) methods to generate and execute redundant trajectories, with PD controllers ensuring error recovery and stability. I also used MATLAB for analysis and validation of the control strategies.

────────────────────────────────────────────────
🔐 Harware Security Analysis of a SoC, 2024
For my bachelor’s thesis, I conducted research on hardware vulnerability analysis of the NRF52810 SoC, focusing on voltage fault injection attacks, under the supervision of Emilio Coppa. During this work, I collaborated with Milena Mangiola, who later extended the research to electromagnetic fault injection, successfully bypassing the chip’s built-in security mechanisms. She published it as CVE-2025-9709.

────────────────────────────────────────────────
🤖 Building a Differential Drive Robot, 2024
As part of the Honours Program at Sapienza University, I developed a differential drive robot under the supervision of Giorgio Grisetti. The project included building the hardware and programming a bare-metal firmware for the ATmega2560 AVR microcontroller. The system receives control commands via UART from a PC-connected joystick to drive the motors with a PID controller, while simultaneously performing odometry-based localization and transmitting position data back to the PC.