Senior Deep Learning Engineer

NVIDIA Cosmos is our open platform of world foundation models for Physical AI, built to interpret images, video, and text and turn them into a structured understanding of a physical scene: motion, object interactions, geometry, and physical context.

The Senior Deep Learning Engineer will own the data engine and end-to-end training verification behind the 3D spatial reasoning and perception capabilities of these models. This involves:

• Sourcing, curating, filtering, and balancing large-scale real-world image and video corpora into vision-language training data
• Building annotation and auto-labeling pipelines for 3D-grounded supervision
• Ensuring data quality end-to-end, including semantic deduplication and automated quality scoring
• Verifying end-to-end model training and running validation pipelines
• Building and operating the 3D and spatial evaluation suite
• Partnering with research scientists to translate 3D research hypotheses into dataset and ablation experiments
• Operating on large multi-node GPU clusters and optimizing data throughput

The ideal candidate has:

• MS or PhD in Computer Science, Electrical/Computer Engineering, Robotics, or a related field
• 12+ years of experience building deep learning systems in Python with PyTorch or JAX on Linux
• Deep expertise in 3D computer vision, multi-view geometry, and structure-from-motion
• Hands-on experience with vision-language models and large-scale multimodal data pipelines
• Experience running and validating large model training on multi-GPU, multi-node clusters
• Rigorous evaluation methodology and excellent written and verbal communication skills