mlops-initialization-cn_skill

Overview

This skill bootstraps a modern Python MLOps project with sensible defaults for packaging, dependency management, linting, type checking, and VS Code. It provides an init script that creates a src/ layout, pyproject.toml configured for uv, VS Code settings, and a ready Git repository. The goal is a reproducible, linted, typed development environment with locked dependencies and standard MLOps conventions. It saves time and enforces consistency across new projects.

How this skill works

Run the provided init script to generate a project skeleton and copy reference configuration files into your new repo. Dependency management uses uv: add packages with uv add, produce a lockfile (uv.lock), and sync environments with uv sync. The repo includes Ruff and MyPy presets wired through pyproject.toml and a .vscode settings file so editors enforce the same rules. The script also initializes Git and populates .gitignore for common Python/MLOps artifacts.

When to use it

• Starting a new machine learning or MLOps project and you want a reproducible setup.
• Onboarding teams that need consistent linting, typing, and editor settings.
• Creating experiment repos that should adhere to production-ready packaging conventions.
• Bootstrapping teaching examples or course materials with a shared project layout.
• Migrating legacy scripts into a structured src/ package layout.

Best practices

• Use the init script as the single source of truth for project structure to maintain consistency.
• Manage dependencies with uv and commit uv.lock to ensure reproducible installs.
• Keep code in src/ to avoid import ambiguity during tests and packaging.
• Enforce static checks (Ruff, MyPy) locally and in CI to catch issues early.
• Include .vscode settings to standardize developer ergonomics, but allow personal overrides in user settings.

Example use cases

• Create a new ML model package with src/, type hints, and CI-ready configs in minutes.
• Set up a lab or workshop repository for students with consistent editor and lint rules.
• Initialize an experiment tracking repo that will later be converted to a deployable package.
• Standardize project templates across a research team to reduce onboarding friction.