Research Scientist, Meta
1 paper at NeurIPS 2025
We propose VisualLens, which leverages task-agnostic user's visual record to enhance the understanding of individual preferences and deliver more personalized MLLM recommendations.