by-deploy-compute

To deploy design tools on your own compute hardware, use a six-field deploy.yaml manifest to automate environment checks, apply per-tool deployment templates, and execute target-specific deployment scripts for local GPUs, RunPod, Modal, or HPC.

Automated tool deployment targets local GPUs, RunPod, Modal, and HPC compute surfaces. Each target requires a valid API key for cloud platforms and uses per-target references to guide the specific deployment configuration and environment setup.

To validate a design tool deployment, run the built-in smoke-test validation after executing the per-target deployment scripts. This ensures operability and automatically patches the config.json file to guarantee the deployed tool is discoverable.

Yes, you need a six-field deploy.yaml manifest to deploy tools to RunPod or Modal. The manifest works alongside an environment check script and per-target references to automate cloning, weights management, and environment setup across cloud targets.

Yes, you can deploy multiple design tools across different compute surfaces using a single six-field deploy.yaml manifest. Per-tool deployment templates handle tool-specific cloning and weights management, while per-target references manage the compute surface configuration.

Automated compute deployment scripts require the pyyaml and requests dependencies. You also need the six-field deploy.yaml manifest, the environment check script, per-target references, and a valid API key for any cloud targets like RunPod or Modal.