Portrait of Dounia Shaaban Kabakibo

Dounia Shaaban Kabakibo

Lab Representative
PhD
Supervisor
Co-supervisor
Research Topics
AI for Science
Deep Learning
Molecular Modeling

Biography

I’m a PhD student in physics at Université de Montréal, co-supervised by Michel Côté and Alex Hernandez Garcia. My interests have wandered from theoretical to computational physics and, more recently, to AI—especially generative models and graph neural networks—which I see as new and exciting tools in my physicist’s toolbox. I use them to explore topics from Raman spectra to battery materials and critical phenomena. Outside of science, I’m passionate about social justice, art, and learning new things.

Since joining Mila, I fell for its vibrant, welcoming community. I got involved in many (mostly unofficial) ways—the talent show, speed science competitions, the mending workshop, and student–administration discussions—until it felt natural to make things official and propose my candidacy as a LabRep. I’m grateful to the students who said yes. As a LabRep, I hope to be able to give back by amplifying student voices and supporting social events that bring people together.

I’m often told I’m a good listener—feel free to reach out if you need an ear or have ideas for us LabReps!

Publications

A Comparative Study of Molecular Dynamics Approaches for Simulating Ionic Conductivity in Solid Lithium Electrolytes
Accurate prediction of ionic conductivity is critical for the design of highperformance solid-state electrolytes in next-generation batterie… (see more)s. We benchmark molecular dynamics (MD) approaches for computing ionic conductivity in 21 lithium solid electrolytes for which experimental ionic conductivity has been previously reported in the literature. Specifically, we compare simulations driven by density functional theory (DFT) and by universal machine-learning interatomic potentials (uMLIPs), namely a MACE foundation model. Our results suggest comparable performance between DFT and MACE, with MACE requiring only a fraction of the computational cost. The framework developed here is designed to enable systematic comparisons with additional uMLIPs and fine-tuned models in future work.