Lightly AG

Lightly AG builds data-centric AI infrastructure for computer vision. Its tools help machine learning teams curate, label and manage visual datasets at scale, with the aim of reducing the cost and time required to train production-ready models. The company is an ETH Zurich spin-off with research roots and is backed by Y Combinator, serving customers globally across autonomous driving, medical imaging and industrial inspection.

The product line spans both commercial and open-source offerings. LightlyOne identifies the most valuable data samples for labeling in order to optimise dataset curation at scale. LightlyTrain provides self-supervised pretraining frameworks intended to remove the need for expensive manual annotation. On the open-source side, LightlySSL is a framework for self-supervised learning, and LabelFormat is a standardised format for data annotation.

Technical work concentrates on self-supervised learning, representation learning, and data curation and selection, applied to machine learning infrastructure for computer vision. The company combines research with practical deployment tooling and contributes to the open-source community.