I build vision-language models and multimodal representations that power search, retrieval, and content understanding across Adobe's creative products. My work also spans visual fingerprinting and content authenticity. Previously, I completed my PhD at the University of Bern, supervised by Paolo Favaro.
Developing multimodal embedding models for visual search, retrieval, aesthetic prediction, and content recommendation. Applications include spatial understanding for intelligent reframing and content repurposing workflows.
Embeddings · VLMs · Retrieval · ReframingRobust image and video fingerprinting applied to content provenance, training data deduplication, asset management, and platform moderation at scale.
Fingerprinting · Deduplication · ProvenanceTraining strategies for learning visual and multimodal representations without human annotation, from images, video, and audio.
Contrastive · Video · AudioFaculty Prize, University of Bern
CVMP 2021
CVPR 2019 (top 1%), ECCV 2020, ECCV 2022
PRAIRIE/MIAI AI Summer School
Joint Alumni Association in Computer Science