CosmicAC Jobs

A job is a unit of work that CosmicAC runs on your Kubernetes cluster. When you create a job, CosmicAC provisions the resources that the job needs, runs the workload on one or more GPUs in your cluster, and tracks the job's status.

Job types

CosmicAC runs two types of jobs. Both run inside a KubeVirt virtual machine instance (VMI) with one or more GPUs, but you work with each one differently:

• GPU Container Job — gives you shell access to the VMI, so you can use it like a remote machine and run your own code. See GPU Container Job.
• Managed Inference Job — serves an open-source model behind an OpenAI-compatible API, using the vLLM runtime for language models or the Parakeet runtime for speech-to-text. See Managed Inference Job.

Job status

A job moves through a sequence of statuses from the moment you create it. Each status reflects the job's current state on your cluster.

The statuses have the following meanings:

• Creating — the job's configuration and GPU requirements are being validated.
• Queued — the GPUs that the job needs are being allocated on your cluster.
• Starting — the job's image is being pulled and the job is being set up. For inference jobs, this includes loading the model weights, which can take several minutes.
• Running — the job is up and accessible. A GPU Container Job provides shell access, and a Managed Inference endpoint accepts requests.
• Restarting — the job's VMI is being replaced, keeping its storage and resources.
• Failed — the job failed to start or crashed.

Managing a job

After you create a job, you can restart or delete it:

• Restart — replaces the job's VMI but keeps its storage and resources.
• Delete — removes the job's VMI, resources, and storage.

Both cosmicac-cli and the web interface support these actions. See CLI commands for jobs.

Next steps

• GPU Container Job
• Managed Inference Job
• Architecture