Francesco Vigni

I’m a research engineer with deep expertise in self-supervised visual learning, video understanding, and perception systems for autonomous agents. My work spans robotics, medical AI, and embedded vision, with a focus on Vision Transformer foundation models trained on large-scale video data.

In my current role, I design SSL pretraining strategies using teacher-student self-distillation and masked modelling, define multi-objective pretraining and fine-tuning protocols, and build benchmarking frameworks to evaluate world-model understanding and prediction. I’m motivated by turning research into innovation—integrating intelligent systems into products and processes that improve everyday life.