Ph.D. candidate at the Federal University of São Paulo’s Institute of Science and Technology (ICT-UNIFESP). Greatly enjoys challenging himself in practical Data Science Challenges from time to time, sometimes to a fault. While researching GNNs, stumbled upon ITU’s challenge and the unfamiliar territory of Queueing Theory and traffic flow in 5G networks. With some perseverance and good fortune, ended up as the Silver Champion of said challenge. During his master’s, investigated the effect of label noise on classic graph-based classifiers and proposed a way to assess the reliability of a label by removing the diagonal of the propagation matrix, evaluating whether a given label could be well estimated by the model itself and all remaining labels. Currently exploring ways to extend this method to make the most out of the label information and optimize parameters in a robust manner within graph neural networks.