feat!: migrate to canonical stack — v5.0.0 (Python 3.14, MLflow 3)

.cruft.json

@@ -11,8 +11,8 @@
       "license": "MIT",
       "version": "4.0.0",
       "description": "Predict the number of bikes available",
-      "python_version": "3.13",
-      "mlflow_version": "2.20.3",
+      "python_version": "3.14",
+      "mlflow_version": "3.14.0",
       "_template": "https://github.com/fmind/cookiecutter-mlops-package"
     }
   },

.dockerignore

@@ -0,0 +1,18 @@
+.git
+.github
+.venv
+.pytest_cache
+.ruff_cache
+.ty_cache
+__pycache__
+*.py[co]
+.coverage
+htmlcov
+dist
+docs
+mlruns
+outputs
+notebooks
+tests
+.env
+.cruft.json

.gemini/skills/MLOps Automation/SKILL.md

@@ -24,10 +24,10 @@ Replace manual commands with a `justfile`.
 1. **Tool**: `just` (modern alternative to Make).
 2. **Organization**: Split tasks into `tasks/*.just` modules (e.g., `tasks/check.just`, `tasks/docker.just`).
 3. **Core Tasks**:
-    - `check`: Run all linters and tests.
-    - `package`: Build wheels.
-    - `clean`: Remove artifacts.
-    - `install`: Setup dev environment.
+   - `check`: Run all linters and tests.
+   - `package`: Build wheels.
+   - `clean`: Remove artifacts.
+   - `install`: Setup dev environment.
 
 ### 2. Pre-Commit Hooks
 
@@ -45,8 +45,8 @@ Reproducibility anywhere.
 1. **Tool**: `docker`.
 2. **Base Image**: Use `ghcr.io/astral-sh/uv:python3.1X-bookworm-slim` for minimal size.
 3. **Optimization**:
-    - **Layer Caching**: Copy `uv.lock` + `pyproject.toml` and run `uv sync` *before* copying `src/`.
-    - **Multi-stage**: Build inputs in one stage, copy only artifacts (`dist/*.whl`) to the runtime stage.
+   - **Layer Caching**: Copy `uv.lock` + `pyproject.toml` and run `uv sync` _before_ copying `src/`.
+   - **Multi-stage**: Build inputs in one stage, copy only artifacts (`dist/*.whl`) to the runtime stage.
 4. **Registry**: ask for the company artifact registry, or use `ghcr.io` for GitHub.
 
 ### 4. CI/CD Workflows
@@ -55,8 +55,8 @@ Automate verification and release.
 
 1. **Platform**: ask for the company CI/CD platform, or use `github-actions` for GitHub.
 2. **Workflows**:
-    - `check.yml`: On PRs (Run `just check`).
-    - `publish.yml`: On Release (Build docker image, publish docs/package).
+   - `check.yml`: On PRs (Run `just check`).
+   - `publish.yml`: On Release (Build docker image, publish docs/package).
 3. **Optimization**: Use `concurrency` to cancel redundant runs.
 
 ### 5. AI/ML Experiments & Registry
@@ -65,11 +65,11 @@ Manage the ML lifecycle.
 
 1. **Platform**: `MLflow`.
 2. **Tracking**:
-    - Use `mlflow.autolog()`.
-    - Log metrics, params, and artifacts.
+   - Use `mlflow.autolog()`.
+   - Log metrics, params, and artifacts.
 3. **Registry**:
-    - Register top models manually or via CI.
-    - **Aliases**: Use `@champion` or `@production` for stable deployment pointers. Never rely on moving versions (e.g., `v1` -> `v2`).
+   - Register top models manually or via CI.
+   - **Aliases**: Use `@champion` or `@production` for stable deployment pointers. Never rely on moving versions (e.g., `v1` -> `v2`).
 
 ### 6. Design Patterns
 

.gemini/skills/MLOps Collaboration/SKILL.md

@@ -41,18 +41,18 @@ Eliminate "it works on my machine".
 
 1. **Templates**: Use `cookiecutter` for scaffolding and `cruft update` to keep projects synced.
 2. **Workstations**: Add `.devcontainer/devcontainer.json`.
-    - Define Docker image, extensions, and settings.
-    - Enable GitHub Codespaces support.
+   - Define Docker image, extensions, and settings.
+   - Enable GitHub Codespaces support.
 
 ### 4. Release Management
 
 Ship with confidence.
 
 1. **Versioning**: Follow **SemVer** (MAJOR.MINOR.PATCH) and [Keep a Changelog](https://keepachangelog.com/).
 2. **Workflows**:
-    - **GitHub Flow**: Small teams, continuous delivery (`main` is stable).
-    - **Git Flow**: Scheduled releases (`develop` + `release` branches).
-    - **Forking**: Open source, distributed contributors.
+   - **GitHub Flow**: Small teams, continuous delivery (`main` is stable).
+   - **Git Flow**: Scheduled releases (`develop` + `release` branches).
+   - **Forking**: Open source, distributed contributors.
 3. **Process**: Bump version -> Update Changelog -> Tag -> Release.
 
 ## Self-Correction Checklist

.gemini/skills/MLOps Industrialization/SKILL.md

@@ -23,18 +23,18 @@ Adopt the `src` layout to prevent import errors and separate source from tooling
 
 1. **Directory Tree**:
 
-    ```text
-    my-project/
-    ├── pyproject.toml       # Dependencies & Metadata
-    ├── uv.lock
-    ├── README.md
-    └── src/
-        └── my_package/      # Main package directory
-            ├── __init__.py
-            ├── io/          # Side-effects (Datasets, APIs)
-            ├── domain/      # Pure business logic (Models, Features)
-            └── application/ # Orchestration (Training loops, Inference)
-    ```
+   ```text
+   my-project/
+   ├── pyproject.toml       # Dependencies & Metadata
+   ├── uv.lock
+   ├── README.md
+   └── src/
+       └── my_package/      # Main package directory
+           ├── __init__.py
+           ├── io/          # Side-effects (Datasets, APIs)
+           ├── domain/      # Pure business logic (Models, Features)
+           └── application/ # Orchestration (Training loops, Inference)
+   ```
 
 2. **Configuration**: Use `pyproject.toml` for all build metadata and dependencies.
 
@@ -43,16 +43,16 @@ Adopt the `src` layout to prevent import errors and separate source from tooling
 Balance structure with predictability.
 
 1. **Domain Layer (Pure)**:
-    - **Rule**: Code here must be deterministic and free of side effects (no I/O).
-    - **Use Case**: Feature transformations, Model architecture definitions.
-    - **Style**: Functional (pure functions) or Immutable Objects (dataclasses).
+   - **Rule**: Code here must be deterministic and free of side effects (no I/O).
+   - **Use Case**: Feature transformations, Model architecture definitions.
+   - **Style**: Functional (pure functions) or Immutable Objects (dataclasses).
 2. **I/O Layer (Impure)**:
-    - **Rule**: Isolate external interactions here.
-    - **Use Case**: Loading data from S3, saving models to disk, logging to MLflow.
-    - **Style**: OOP (Classes to manage connections/state).
+   - **Rule**: Isolate external interactions here.
+   - **Use Case**: Loading data from S3, saving models to disk, logging to MLflow.
+   - **Style**: OOP (Classes to manage connections/state).
 3. **Application Layer (Orchestration)**:
-    - **Rule**: Wire Domain and I/O together.
-    - **Use Case**: Tuning, Training, Inference, Evaluation, etc.
+   - **Rule**: Wire Domain and I/O together.
+   - **Use Case**: Tuning, Training, Inference, Evaluation, etc.
 
 ### 3. Application Entrypoints
 
@@ -61,46 +61,46 @@ Create standard, installable CLI tools.
 1. **Define Script**: Create `src/my_package/scripts.py` with a `main()` function.
 2. **Register**: Add to `pyproject.toml`:
 
-    ```toml
-    [project.scripts]
-    my-tool = "my_package.scripts:main"
-    ```
+   ```toml
+   [project.scripts]
+   my-tool = "my_package.scripts:main"
+   ```
 
 3. **CLI Execution**:
-    - **Dev**: `uv run my-tool` (No install needed).
-    - **Prod**: `pip install .` -> `my-tool` (Installed on PATH).
+   - **Dev**: `uv run my-tool` (No install needed).
+   - **Prod**: `pip install .` -> `my-tool` (Installed on PATH).
 4. **Guard**: Always use `if __name__ == "__main__":` in scripts to prevent execution on import.
 
 ### 4. Configuration Management
 
 Decouple settings from code using **OmegaConf** (Parsing) and **Pydantic** (Validation).
 
 1. **Define Schema (Pydantic)**:
-    - Create a class that defines *expected* types and defaults.
+   - Create a class that defines _expected_ types and defaults.
 
-    ```python
-    from pydantic import BaseModel
+   ```python
+   from pydantic import BaseModel
 
-    class TrainingConfig(BaseModel):
-        batch_size: int = 32
-        learning_rate: float = 0.001
-        use_gpu: bool = False
-    ```
+   class TrainingConfig(BaseModel):
+       batch_size: int = 32
+       learning_rate: float = 0.001
+       use_gpu: bool = False
+   ```
 
 2. **Parse & Validate (OmegaConf)**:
-    - Load YAML, merge with CLI args, and validate against the schema.
+   - Load YAML, merge with CLI args, and validate against the schema.
 
-    ```python
-    import omegaconf
+   ```python
+   import omegaconf
 
-    # 1. Load YAML
-    conf = omegaconf.OmegaConf.load("config.yaml")
-    # 2. Merge with CLI (optional)
-    cli_conf = omegaconf.OmegaConf.from_cli()
-    merged = omegaconf.OmegaConf.merge(conf, cli_conf)
-    # 3. Validate -> Returns a validated Pydantic object
-    cfg: TrainingConfig = TrainingConfig(**omegaconf.OmegaConf.to_container(merged))
-    ```
+   # 1. Load YAML
+   conf = omegaconf.OmegaConf.load("config.yaml")
+   # 2. Merge with CLI (optional)
+   cli_conf = omegaconf.OmegaConf.from_cli()
+   merged = omegaconf.OmegaConf.merge(conf, cli_conf)
+   # 3. Validate -> Returns a validated Pydantic object
+   cfg: TrainingConfig = TrainingConfig(**omegaconf.OmegaConf.to_container(merged))
+   ```
 
 3. **Secrets**: Use Environment Variables (`os.getenv`), never commit them.
 
@@ -110,18 +110,18 @@ Make code usable and maintainable.
 
 1. **Docstrings**: Use **Google Style** docstrings for all modules, classes, and functions.
 
-    ```python
-    def calculate_metric(y_true: np.ndarray, y_pred: np.ndarray) -> float:
-        """Calculates the accuracy score.
+   ```python
+   def calculate_metric(y_true: np.ndarray, y_pred: np.ndarray) -> float:
+       """Calculates the accuracy score.
 
-        Args:
-            y_true: Ground truth labels.
-            y_pred: Predicted labels.
+       Args:
+           y_true: Ground truth labels.
+           y_pred: Predicted labels.
 
-        Returns:
-            The accuracy as a float between 0 and 1.
-        """
-    ```
+       Returns:
+           The accuracy as a float between 0 and 1.
+       """
+   ```
 
 2. **Type Hints**: Use standard python typing (`typing`, `list[str]`) everywhere.