Helix AI Pretraining Engineer for Multimodal Foundations

Figure; San Jose (CA); On-site; USD 120,000 - 150,000; Full time

Job summary

A leading AI robotics company in San Jose seeks a Helix AI Engineer, Pretraining, to build large-scale foundation models for humanoid autonomy. Ideal candidates will have experience in training models, a strong grasp of deep learning architectures, and proficiency in Python, especially with PyTorch. This role involves designing evaluation frameworks, developing pretraining strategies, and collaborating with various teams. The position requires in-office collaboration 5 days a week.

Qualifications

• Experience training large-scale foundation models or working on pretraining for LLMs or multimodal systems.
• Strong understanding of modern deep learning architectures.
• Experience with large-scale distributed training and optimization.
• Proficiency in Python and deep learning frameworks.
• Strong experimental rigor and ability to iterate on model design.
• Solid software engineering skills and ability to build scalable systems.
• Ability to operate independently on high-impact problems.

Responsibilities

• Design and train large-scale foundation models across multimodal data.
• Develop pretraining strategies that improve generalization and reasoning.
• Explore architectures including transformer-based and emerging paradigms.
• Build large-scale distributed training pipelines across GPU clusters.
• Collaborate to integrate pretrained models into the autonomy stack.
• Design evaluation frameworks for reasoning ability and robustness.
• Contribute to post-training approaches including fine-tuning.

Skills

Experience training large-scale foundation models

Understanding of modern deep learning architectures

Experience with large-scale distributed training

Proficiency in Python

Strong experimental rigor

Solid software engineering skills

Ability to drive high-impact technical problems

Tools

PyTorch

Job description

A leading AI robotics company in San Jose seeks a Helix AI Engineer, Pretraining, to build large-scale foundation models for humanoid autonomy. Ideal candidates will have experience in training models, a strong grasp of deep learning architectures, and proficiency in Python, especially with PyTorch. This role involves designing evaluation frameworks, developing pretraining strategies, and collaborating with various teams. The position requires in-office collaboration 5 days a week.