Founding Research Engineer: World Models & Simulation

Bonfirevc; Palo Alto (CA); On-site; USD 170,000 - 230,000; Full time

Job summary

Orbifold AI is building the infrastructure layer for Physical AI in Palo Alto, CA. The research engineer will develop world models and evaluation environments to test robotic policies at scale.

You will train models on real captured data, generate synthetic data, and build a robust rollout harness. A PhD with first-author work in world models or related areas is highly valued, with strong ML engineering skills a must.

Qualifications

• PhD or equivalent research experience with first-author publications or shipped work in world models and related fields.
• Strong ML engineering with distributed training experience.
• Clear-eyed view on data usefulness and model limitations.
• Comfortable being measured on real-world policy improvements.

Responsibilities

• Train generative world models on multi-view real-world data and evaluate policies.
• Generate targeted synthetic data to address measured deficits.
• Build rollout and evaluation harness with scenar io suites and reproducible runs.
• Measure the reality gap rigorously and drive data collection efforts.
• Scale training with multi-node PyTorch on Ray.

Skills

World models

Action-conditioned video

Video diffusion

Model-based RL

Neural rendering

Robot learning in simulation

Education

PhD or equivalent

Tools

PyTorch

Ray

MuJoCo

Isaac Sim

Genesis

ManiSkill

Job description

Orbifold AI is building the infrastructure layer for Physical AI in Palo Alto, CA. The research engineer will develop world models and evaluation environments to test robotic policies at scale.

You will train models on real captured data, generate synthetic data, and build a robust rollout harness. A PhD with first-author work in world models or related areas is highly valued, with strong ML engineering skills a must.