[...]

Builds ml-api using ml-api/Dockerfile

Expects model code, dependencies, Dockerfile, and fine-tuned assets to live under ml-api/

References ml-api via relative paths (for example, in local_platform/builder.py)

Removing the directory and keeping only a Git URL would require adding cloning and synchronization logic before every build. By keeping ml-api in the repository, the control plane can:

Operate offline

Rebuild quickly without extra setup

Version infrastructure, model code, and assets together

To update ml-api, simply pull upstream changes into the ml-api/ directory and re-run the deployment.

Learn more: see the control plane build flow and path assumptions in local_platform/builder.py.

Extending this workspace

If you decide to extend the local-first stack, feel free to reach out before making structural changes; we can coordinate so both ml-api/ and local_platform/ continue to work together. The control plane expects the artifacts listed above, so keep those paths intact when you test new builds or add services.