Research Scientist - World Model
Remote • Redwood City, CA • FullTime
Posted 2mo ago
Job Location
Redwood City, CA
Tech Stack
Remote Work Policy
Fully remote
Employment Type
FullTime
Categories
AI Research Engineer
About the job
Luma is seeking a Research Scientist to transform their industry-leading generative video models into interactive, controllable, and physically faithful world models. This role is central to the company's mission, focusing on inventing next-generation world-model architectures and developing the controllability mechanisms that enable agents to operate within these generated environments. You will define and own the key metrics for success, conduct scaling studies, and contribute to open-source releases. This is a broad and open-ended research problem for individuals with deep expertise in generative modeling or model-based reinforcement learning.
Responsibilities
- Invent next-generation world-model architectures (diffusion, transformer, autoregressive, or hybrid), focused on controllability and physical consistency.
- Develop controllability mechanisms such as action conditioning, view conditioning, and long-horizon rollouts to enable agent interaction.
- Define and own metrics for physical fidelity, long-horizon coherence, action-following, and downstream usefulness for policy training.
- Conduct scaling studies to identify optimal configurations of compute, data, and architecture.
- Publish research at the frontier of the field and contribute to open-source releases.
Requirements
- PhD or equivalent research record in ML, computer vision, robotics, or a related field.
- Deep expertise in large-scale generative modeling (video/3D/world), self-supervised representation learning, or model-based RL.
- Strong PyTorch and large-scale training experience, capable of training models on multi-node clusters.
- A recognized research record, demonstrated through top-venue publications and/or widely used open releases.