What does Macrodata Labs do?
Macrodata Labs builds data-processing infrastructure for physical-AI training, turning robot and egocentric video recordings into training-ready datasets with timestamped subtask labels, 3D hand trajectories, and task-progress signals. Founded in 2026 by Guilherme Penedo and Hynek Kydlíček, former Hugging Face researchers behind the FineWeb and FinePDFs datasets, the Paris-based company also maintains Refiner, an open-source robotics data processing framework.