Valter Hudovernik is an M.Sc. student in Data Science at the University of Ljubljana. His thesis, Deep Learning Methods for Synthetic Relational Data Generation, explores the use of diffusion models to generate private, high-fidelity synthetic databases. He presented preliminary results at the NeurIPS 2024 Tabular Representation Learning Workshop and is finalizing a publication on benchmarking synthetic relational data.

Alongside his studies, Valter has gained hands-on experience as a student researcher at the Jožef Štefan Institute and a data science intern at Endava. He also collaborated on an interdisciplinary project at the Geological Survey of Slovenia, developing explainable models for predicting attributes of karst water sources.

Beyond data science, he holds a black belt in Brazilian Jiu-Jitsu and received the University Award for Special Achievements for his competition success. His passion for sports and machine learning converged in his bachelor’s thesis, which earned him the Faculty Prešeren Award.