This is a fully remote position, open to applicants in California.

📋 Description

• Design and refine our VLA model architecture, which includes the VLM backbone, action decoder, and multimodal fusion pipeline.

• Construct and enhance large-scale training infrastructure, covering distributed training, data pipelines, mixed-precision, and efficient fine-tuning.

• Create simulation-based evaluation and closed-loop training workflows utilizing photorealistic neural rendering.

• Curate and oversee multimodal training datasets that encompass both real-world driving and synthetic scenarios.

• Convert cutting-edge research (such as diffusion/flow-matching action heads, reasoning-augmented VLAs, and world models) into production-ready systems.

• Work closely with vehicle systems and controls engineers to incorporate model outputs into a real-time autonomous driving framework.

⛳️ Requirements

• MS or PhD in Computer Science, Machine Learning, Robotics, or a related field, or equivalent industry experience.

• Strong expertise in PyTorch, distributed training, and GPU-accelerated workflows.

• Robust understanding of transformer architectures, attention mechanisms, and contemporary generative modeling (including diffusion and flow matching).

• Must be eligible to work in the United States.