Experience

Visiting researcher
Mila, with David Rolnick and Yoshua Bengio
Feb. 2022 - 2023

• Designed scalable and expressive symmetry-preserving GNNs to predict the relaxed adsorption energy of a catalyst-adsorbate system. Combine it with a generative method (e.g. GFlowNet) for catalyst discovery.

Research intern
Inria Saclay, with Fragkiskos Malliaros
June - Dec. 2020

• Proposed a unified view of existing GNN explainability methods along with a new explainer.
• Investigated graph representation learning, explainability and fusion of protein-protein interaction nets.

Student Researcher
Naver Labs Europe, with Matthias Gallé
March - June 2020

• Worked on controlled and contextualised text generation for novel authors. Open-source writing assistant.