Member of Technical Staff, Foundation Models

The role

You will build and train large foundation world models from scratch. This involves working across the full ML stack including data, architecture, training, inference, and evaluation.

What you will do

• Learn what makes large real-time world models tick.
• Run scaling studies to fit scaling laws over model size, data, and compute.
• Own world model training at scale across thousands of GPUs.
• Implement state of the art ML algorithms and define metrics.
• Work end to end across the stack from data pipelines to real-time inference.
• Exploit latest features on modern GPUs to increase efficiency.
• Take ownership of the full ML stack and core frameworks.

Required qualifications

• 2+ years of software engineering experience, with significant work in ML performance.
• 4+ years of ML engineering experience, or a PhD in a related field.
• Hands-on experience training large models and pretraining at multi-node scale.
• Comfortable reasoning about scale, scaling laws, compute and data budgets.
• Track record of owning projects end to end.
• Proficiency with PyTorch (or TF/JAX).

About the company

Odyssey is an AI lab pioneering general world models: causal, multimodal systems that learn to predict and interact with the world over long horizons.