workbench-reference-workflows

You can orchestrate physical-AI workflows by using declarative SkyPilot YAML templates to define resource requirements and execution order. This skill provides validated templates and CLI command structures to launch complex simulation and training jobs on Nebius GPU clusters without re-implementing the underlying DAG.

Declarative orchestration for physical-AI workflows uses SkyPilot YAML templates to standardize the definition of resource requirements, environment configurations, and execution order. This approach ensures consistency across diverse simulation, training, and evaluation tasks on Nebius infrastructure.

Running a parameter sweep for an Isaac Lab RL job requires a validated YAML template and CLI command structure provided by this skill. It allows you to launch the job on Nebius GPU clusters using a standardized interface for diverse robotics and VLM research pipelines.

Yes, executing VLM evaluation tasks requires adherence to a three-tier contract involving CLI submission, SDK helpers, and declarative SkyPilot YAML definitions. This standardized execution interface ensures consistent management of diverse physical-AI workflows.

The best way to manage diverse physical-AI pipelines is using a standardized, declarative approach to workflow orchestration. Utilizing SkyPilot YAML templates ensures consistency across simulation, training, and evaluation tasks on Nebius infrastructure without re-implementing the underlying DAG.