project-scaffold

To generate an ML project scaffold, you can run a scaffold generator on an experiment-plan.md to automatically produce a src/ tree, pyproject.toml, and Hydra configs. This removes manual boilerplate work and creates a runnable project structure ready for immediate coding.

In ML project structure, Hydra config provides dynamic configuration management, while Factory/Registry wiring implements a pattern to dynamically instantiate objects. Using these together standardizes experiment execution and ensures the generated scaffold remains modular and easily extendable.

Yes, you can apply the scaffold generator at project inception without a detailed experiment plan. It produces standard dependency templates, an entry point, and a test skeleton, giving you a versioned project layout to start coding against immediately.

Yes, the generated ML scaffold includes a test skeleton for validating code and a Makefile for automating common development commands. It also provides a pyproject.toml for dependency management, ensuring a complete and runnable environment from the start.

Using a scaffold generator instead of manually setting up ML project boilerplate ensures consistent project structure and saves time. It automatically infers dependencies from your experiment plan and generates standardized Factory/Registry wiring and Hydra config templates.

The best way to structure a runnable ML project is to generate a standardized scaffold with a src/ tree, entry point, and Makefile. This versioned layout is specifically designed to be ready for downstream skills like experiment-data-builder and model-setup.