PhD in Smart Computing
"Random Access Memories"
I am a PhD student in Smart Computing and Artificial Intelligence at the University of Florence, working in collaboration with the University of Siena. My current research focus is fake and synthetic-content detection, especially for images: I study which representations preserve useful traces of the generation process and how those traces behave across models, datasets, and common image transformations.
A package for extracting hidden states and attention maps from Stable Diffusion 3.5 MM-DiT models for AI-image detection, attribution, and interpretability experiments.
An experimental decoder-only Transformer for Dante-style text generation, exported for browser use through ONNX Runtime Web.
A Python command wrapper that updates a Telegram dashboard with recent output, resource information, and controls for a running command.
A conditional autoencoder for exploring chemical representations relevant to the early-stage screening of pharmaceutical contaminants for adsorption by phenyl-functionalized poly(2-oxazoline).
A Python reimplementation of mshibanami/GitHubTrendingRSS that generates Atom feeds for GitHub Trending views and a simple static HTML index.
A study of density estimation from unlabeled samples, combining estimates from maximum-likelihood Gaussian mixture models with the generalization of an artificial neural network.
A Python and PyTorch GAN course project that generates images inspired by Lissajous curves and harmonographs from a synthetic dataset.
An ESANN 2022 comparison of DeepLab V2 and MobileNet approaches for segmenting glomeruli in histological whole-slide images from mouse and rat tissue.
A Python Telegram bot, substantially rewritten from RobotRSS, for saving RSS URLs, fetching feeds, and sharing converted links through Telegram.
A collaboration with Gabriel Volpi that turns visual material related to biomedical segmentation and network artifacts into a small generative-art gallery.
A PHP barcode and database manager built for a Computer Networks course project, with to-do and warehouse lists.
