Razvan Pascanu

Member, AIRi@UTCN Advisory Committee

Senior Research Scientist, Google DeepMind | Affiliate Member, Mila – Quebec Artificial Intelligence Institute | General Chair, NeurIPS 2026

Răzvan Pașcanu is a researcher in artificial intelligence and machine learning, specialising in deep learning, neural network optimisation, recurrent neural networks and continual learning. He is a Senior Research Scientist at Google DeepMind and an Affiliate Member of Mila – Quebec Artificial Intelligence Institute. As a member of the AIRi@UTCN Advisory Committee, he contributes his scientific expertise and international perspective to the institute's strategic development, research priorities and international collaborations.

Originally from Romania, Pașcanu pursued his university studies in Germany, earning his master's degree from Jacobs University Bremen in 2009. He completed his PhD at Université de Montréal in 2014 under the supervision of Professor Yoshua Bengio, focusing on optimisation, recurrent neural networks and deep learning.

His research spans several areas of machine learning, including efficient optimisation and training of deep neural networks, memory and recurrent architectures, reinforcement learning, meta-learning, continual learning and graph neural networks. He also contributed to the development of Theano, an influential open-source framework for deep learning research.

Alongside his scientific work, Pașcanu has taken on leadership roles in major international AI conferences. He serves as General Chair of NeurIPS 2026, one of the world's leading conferences in machine learning, following his appointment as Program Chair for NeurIPS 2025. His academic service also includes organisational roles in the Conference on Lifelong Learning Agents (CoLLAs), the Learning on Graphs (LoG) Conference and research workshops dedicated to continual learning.

Beyond his research, Pașcanu is actively involved in strengthening AI education and research communities, particularly in Eastern Europe. He is one of the organisers of the Eastern European Machine Learning Summer School (EEML), an initiative that provides advanced machine learning education and connects emerging researchers with leading international experts. He is also involved in AI Romania and has helped organise Romanian AI Days since 2020, as well as educational activities introducing artificial intelligence to high school students.

Through his involvement in AIRi@UTCN, Răzvan Pașcanu brings expertise in advanced machine learning research, international scientific collaboration and the development of AI research and education communities.