About

Hello there! I’m Vale, and this is my blog. Here is where I collect notes on the AI/ML ideas I find fascinating. Most grow out of my research; some are just curiosity.

I’m a researcher at Ericsson, exploring how AI/ML can power the next generation of video compression technologies.

Before that, I’ve had the privilege of working as a research assistant at the Technical University of Munich , specifically at the Chair of Signal Processing (MSV). There, I developed deep learning models based on transformers and graph neural networks for the physical layer of wireless communications. I also studied cross-domain generalization in neural networks, using kernel-based and statistical methods to explain robust transfer across distribution shifts without target-domain data. This research culminated in my Dr.-Ing. summa cum laude from TUM in 2025 (see dissertation).

During my doctorate, I collaborated with Futurewei Technologies (USA) and Imperial College London to develop AI-driven solutions for next-generation wireless systems. At the same time, I supported the Numerical Linear Algebra for Signal Processing course in the Master’s program by conducting tutorials (exercise lectures) and preparing programming projects.

I hold a BSc and an MSc (both with honors) in Telecommunication Engineering from Politecnico di Milano. During my master’s studies, I did an exchange and wrote my master’s thesis at TUM, while also working as a signal processing student at Intel Corporation during the “5G era”.

I’m a big fan of Weights & Biases (wandb), an awesome tool for tracking ML experiments. Feel free to check my ML portfolio there!