ML Pipeline Expert
Guides building ML training pipelines with experiment tracking, orchestration, and validation
- Repository
- Jeffallan/claude-skills
- Stars
- 12,176
- Licence
- MIT
- Files
- 6 files
What it does
When installed, the agent works as a senior machine-learning infrastructure engineer. It helps design pipelines that move data through validation, feature engineering, training, experiment tracking, and evaluation before a model is promoted. It supplies code starting points for tools such as MLflow, Kubeflow Pipelines, Airflow, Feast and Great Expectations, and follows rules like versioning data and models, logging hyperparameters, and keeping credentials out of code.
Great for
- Engineers setting up MLOps tooling such as MLflow experiment logging or Kubeflow/Airflow training DAGs
- Data science teams who want reproducible training runs and a model registry
- Developers adding data validation checkpoints before model training
Skip it if
- Non-technical users who only need a one-off model or a quick analysis: the skill targets production infrastructure and can add more complexity than needed
- Anyone who would deploy generated pipeline code to production without review: the templates are minimal examples, and the skill itself notes they need error handling, alerting and tests added
- Teams that do not use the named tools (MLflow, Kubeflow, Airflow, Feast, W&B): the guidance is tool-specific
Try asking
- “Set up MLflow experiment tracking for my scikit-learn classifier and register the best model”
- “Create an Airflow DAG that retrains my model weekly with data validation before training”
- “Design a Feast feature store schema for my customer churn features”
- “Add a Great Expectations data check to my training script that halts on failure”
Install in 3 steps
- 1
Download the skill
The ZIP contains the 6 original files, the MIT licence and an attribution note.
ml-pipeline.zip - 2
Add it to your agent
Claude Code (all projects)
mkdir -p ~/.claude/skills && unzip ml-pipeline.zip -d ~/.claude/skills/Claude Code / Agent SDK (this project only)
mkdir -p .claude/skills && unzip ml-pipeline.zip -d .claude/skills/Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.
- 3
Ask for the task
No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.
Get better results
- Give the agent your existing pipeline code and infrastructure tool so it adapts the templates rather than starting from scratch
- Ask for the validation and rollback steps explicitly, since the skill asks for them but the generated code still needs testing
- Review any credentials handling yourself and confirm generated configs use your secrets manager
What's inside
- references/experiment-tracking.md
- references/feature-engineering.md
- references/model-validation.md
- references/pipeline-orchestration.md
- references/training-pipelines.md
- SKILL.md
Redistributed unmodified from Jeffallan/claude-skills under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.