Helix AI Engineer, Video Pretraining

What you'll do

• Design and train video foundation models and pretraining strategies that learn temporal dynamics, motion, and object interaction.
• Build transferable representations, evaluation frameworks, and efficient pipelines for downstream autonomy tasks.
• Integrate pretrained models with generative modeling, agent, and robot learning teams; optimize distributed training performance.

What they ask for

• Experience training large-scale models on video data or other high-dimensional sequential modalities.
• Strong understanding of deep learning architectures for video, vision, or multimodal systems, and large-scale pretraining.
• Proficiency in Python and deep learning frameworks such as PyTorch; experience with distributed training systems and large GPU clusters.
• Strong experimental rigor, software engineering skills, and ability to independently drive ambiguous research directions.

Nice to have

• Experience with frontier video or multimodal foundation models; video diffusion, autoregressive video modeling, or world models.
• Robotics or embodied AI familiarity; large-scale video dataset construction, relevant publications, or experience at a listed leading AI lab.

Pay

$200,000 to $400,000 per year (base salary)

Tools

Python, PyTorch, transformer-based, diffusion-based

AI focus

Pretrains large-scale video foundation models to learn motion, interaction, and temporal structure for perception, prediction, control, and embodied reasoning.