Figure
Helix AI Engineer, Generative AI
About this role
Figure AI is seeking a Helix AI Engineer to develop and deploy large-scale generative models—including diffusion and multimodal approaches—that enable humanoid robots to perceive, reason, and interact with the physical world. The role involves training models for vision, video, and perception tasks, optimizing distributed training pipelines, and integrating generative systems into the full autonomy stack.
What you'll do
- Design, train, and deploy large-scale generative models with focus on diffusion-based approaches for vision, video, and multimodal data
- Develop models that enhance robot perception, world modeling, and prediction from sensory inputs
- Build generative systems for synthetic data creation and dataset scaling for robot learning
- Optimize training pipelines for generative models across distributed systems
- Integrate generative models into the autonomy stack working with data, infrastructure, and agent teams
- Evaluate model quality, robustness, and generalization in real-world robot scenarios
What they're looking for
- Generative modeling (diffusion, autoregressive methods)
- Deep learning for vision and multimodal systems
- Python and PyTorch
- Large-scale distributed training
- Experimental rigor and rapid iteration
- Software engineering and system design
- Independent problem-solving
Opens the application — the Jobs AI extension fills it for you. Set up autofill
Opens the official application on the employer’s site. No login required.
Figure
Figure develops advanced humanoid robots powered by AI technology. The company is hiring engineers across mechanical design, firmware development, manufacturing, quality assurance, and security to build and refine its autonomous robotic systems.
View all jobs at FigureLikely interview questions
- Walk us through a generative model you've trained at scale—what architecture did you use, what were the key challenges, and how did you measure success?
- Describe your experience optimizing training pipelines for large-scale models. What distributed systems have you worked with, and how did you handle bottlenecks?