Autonomy Engineer - VLA Pre-training

What you'll do

• Post-train policies using behavior cloning and reinforcement learning, owning the workflow from data to deployment.
• Shape data collection and work with partners to obtain diverse, high-quality training data.
• Run training on the VLA stack and build data pipelines, while collaborating to scale distributed training and optimize edge inference.

What they ask for

• 3+ years building deep-learning systems, with shipped models or published artifacts.
• Hands-on experience with at least one of LLMs, VLMs, or image/video generative models, including architecture, training, and inference.
• Experience with deep-learning infrastructure, including streaming datasets, checkpointing and state management, and distributed training.
• Strong Python and PyTorch/JAX skills.

Nice to have

• Robotics or autonomous driving experience; experience applying RL to LLMs or robotics.
• Experience with VLA models or productizing deep nets under latency, throughput, telemetry, or on-device constraints.

Pay

$180,000–$300,000 USD per year (base salary)

Tools

Python, PyTorch, JAX, LLMs, VLMs, RL, OpenVLA, Physical Intelligence (π) models

AI focus

Pre-train and fine-tune VLA policies for robots; use behavior cloning, reinforcement learning, synthetic data, and new modalities or architectures.