Helix AI Engineer, Modeling

About this role

Figure AI is seeking an AI Engineer to design and develop core model architectures for humanoid robot autonomy, focusing on multimodal perception, reasoning, and action capabilities. This role involves advancing vision-language-action systems and representation learning approaches that directly enable robot intelligence, with 5 days/week in-office work in San Jose.

What you'll do

• Design model architectures for perception, reasoning, and action across multimodal inputs
• Develop structured world representations including objects, dynamics, and interactions
• Advance multimodal fusion and cross-modal reasoning techniques
• Improve model generalization, robustness, and long-horizon reasoning capabilities
• Collaborate across pretraining, RL, and robot learning teams to integrate advances into autonomy stack
• Design experiments and evaluation frameworks to guide model iteration

What they're looking for

• Deep learning model design and training
• Vision, language, and multimodal systems
• Transformer architectures and modern deep learning approaches
• Python and PyTorch
• Experimental design and rigorous iteration
• Software engineering and system building
• Representation learning and world models
• Embodied AI and robotics fundamentals