AI Engineer, Policy, Optimus

Tesla is scaling the manufacturing of Optimus to the order of millions of bots. The scale of this fleet of bots unlocks previously infeasible methods for real world learning, real2sim2real and policy evaluation. Similar to the end-to-end models for Full Self Driving, we strive for simple but scalable algorithms for Optimus. As an AI Engineer working on Optimus Policy, you will work on the data, model training, evaluation and deployment of Optimus policies for a wide range of tasks – including but not limited to loco-manipulation, dexterous manipulation, navigation outdoors and on uneven terrain etc. You will get to work on the full training stages of the model – from pre-training to reinforcement learning.

Responsibilities

Create and scale datasets from a variety of sources (human, bot, simulation) Design and train efficient and performant policies that complete a wide variety of tasks on Optimus Create evaluation benchmarks for Optimus policies for fast iterative improvement Large-scale pre-training and reinforcement learning to create intelligent and robust policies Deploy policies on Optimus

Requirements

Hands-on experience training and deploying policies for robotics or autonomous systems Strong proficiency in PyTorch and large-scale distributed training Proven track record shipping real-world neural policies Familiarity with multimodal sensing (vision, proprioception, tactile) and modern ML methods Production engineering mindset focused on safety-critical, low-latency systems Passion for building general embodied intelligence through reinforcement learning