Two papers accepted at CVPR 2025: on context-aware multimodal pretraining and continual multimodal pretraining via model-merging over time.
Two papers accepted at NeurIPS 2024: a large-scale study on continual multimodal pretraining and reward-optimized image generation via ReNO.
One paper accepted at ECCV 2024: on intervention-aware concept bottleneck models (supervised master’s thesis project).
One paper accepted at CoLLAs 2024: a critical investigation of rehearsal-free continual learning with pretrained models.
Two main-conference papers accepted at ICLR 2024 (one spotlight): on general knowledge transfer and training-free interpretable compositional image retrieval; plus one workshop paper.
Paper accepted at ICCV 2023: 'WaffleCLIP' studying LLM descriptions for open-vocabulary image classification.
Recipient of the Qualcomm Innovation Fellowship Europe 2023.
Served as reviewer for top conferences (CVPR, ECCV, ICCV, NeurIPS, ICLR, ICML) and journals (TPAMI, MICCAI); recognized as Outstanding Reviewer for CVPR 2022/2023/2025, ECCV 2022/2024, ICCV 2025, ICLR 2024, and NeurIPS 2024.