VLAAD
A lightweight collision-aware plug-in that learns temporally localized risk representations from captioned driving video.
Project detailsI’m a PhD researcher at McGill University working on machine learning for reliable visual systems.
My current work spans vision-language learning, anomaly detection, and the structure of pretrained representations. I’m especially interested in methods that remain useful when labels are weak, anomalies are rare, or deployment data shifts.
I’m currently interested in research internships for 2027.
Selected work
View all
Fitting choices shape the predictive cost of exact group models
Testing whether the predictive penalty of exact group models comes from the group laws—or from how the models are fitted.
Project details
Unveiling the Flaws
A critical analysis of how window size, random seeds, and normalization can dominate reported anomaly-detection performance.
Project details