VLA Pre-training Engineer (Deep Learning)
This role involves developing and training deep learning models for vision-language-action (VLA) systems used in humanoid robotics. The engineer will work on the full lifecycle from data curation and synthetic data generation to distributed training, model optimization, and deployment. Close collaboration with data and MLOps teams ensures scalable, real-time inference on robotic platforms, with a strong focus on improving policy performance through behaviour cloning and reinforcement learning.