Nov 2026 Nov 2026
Arthur Bianchessi and Marcelo Mussi selected as 2026 Google PhD Fellows Arthur Bianchessi e Marcelo Mussi são selecionados como 2026 Google PhD Fellows
2026 Google PhD Fellowship · With support from Google.org · MALTA, PUCRS 2026 Google PhD Fellowship · Com apoio do Google.org · MALTA, PUCRS
Thrilled to see two of my advisees selected as 2026 Google PhD Fellows! Arthur Bianchessi and Marcelo Mussi, Ph.D. students in Computer Science at PUCRS and members of MALTA (Machine Learning Theory and Applications Lab), are recipients of the 2026 Google PhD Fellowship, with support from Google.org.
Arthur studies how Transformers represent position and how they generalize to contexts longer than those seen in training. He is the first author of Bayesian Attention Mechanism: A Probabilistic Framework for Positional Encoding and Context Length Extrapolation, presented at ICLR 2026.
Marcelo works on misinformation identification with LLMs, within the Misinformation Identification with LLMs project. He is the first author of Automated Fact-Checking in Brazilian Portuguese: Resources and Baselines (STIL 2025) and a co-author of Latent Fact-Checking: Detecting Misinformation through Activation Engineering, Best Paper at BRACIS 2026.
They join a cohort of 148 new Fellows from 38 countries, supported by $13 million in funding from Google.org. Each Fellow receives funding for their doctoral research and is connected with a lifelong peer community. Since 2009, the Google PhD Fellowship Program has supported more than 1,100 graduate students worldwide. The full list of recipients is at goo.gle/phd-fellows-2026.
Congratulations, Arthur and Marcelo!
Muito feliz em ver dois orientandos meus selecionados como 2026 Google PhD Fellows! Arthur Bianchessi e Marcelo Mussi, doutorandos em Ciência da Computação na PUCRS e integrantes do MALTA (Machine Learning Theory and Applications Lab), foram selecionados para o 2026 Google PhD Fellowship, com apoio do Google.org.
O Arthur estuda como Transformers representam posição e como generalizam para contextos mais longos do que os vistos no treinamento. Ele é primeiro autor de Bayesian Attention Mechanism: A Probabilistic Framework for Positional Encoding and Context Length Extrapolation, apresentado na ICLR 2026.
O Marcelo pesquisa identificação de desinformação com LLMs, dentro do projeto Identificação de Desinformação com LLMs. Ele é primeiro autor de Automated Fact-Checking in Brazilian Portuguese: Resources and Baselines (STIL 2025) e coautor de Latent Fact-Checking: Detecting Misinformation through Activation Engineering, Best Paper no BRACIS 2026.
Eles fazem parte de um grupo de 148 novos Fellows de 38 países, apoiados por US$ 13 milhões em financiamento do Google.org. Cada Fellow recebe financiamento para sua pesquisa de doutorado e passa a integrar uma comunidade global de pares. Desde 2009, o Google PhD Fellowship Program já apoiou mais de 1.100 estudantes de pós-graduação no mundo. A lista completa de selecionados está em goo.gle/phd-fellows-2026.
Parabéns, Arthur e Marcelo!
See all 2026 Google PhD Fellows → Veja todos os 2026 Google PhD Fellows →