Hi I’m Paolo, a research intern at Gladia and MSc Computer Science student at Sapienza University of Rome, working in mechanistic interpretability and adversarial attacks on diffusion models. I earned my Bachelor’s in Computer Science from the University of Tor Vergata, focusing on algorithms and theoretical computer science. During my Master’s at Sapienza University of Rome, I initially planned to continue in algorithms, but an early exposure to deep learning completely got me hooked and instantly shifted my focus. I became deeply interested in mechanistic interpretability and the core principles of how deep learning models work, which led me to join the Gladia Lab, an active research group in these areas.