How does this handle data versioning and lineage?

It implements best practices for data lineage and versioning using tools like DVC, ensuring every model run is reproducible and traceable to specific data versions.

Which orchestration tools does this skill support?

The skill provides patterns and templates for popular orchestration tools including Apache Airflow, Dagster, Kubeflow Pipelines, and Prefect.

Does it include support for experiment tracking?

Yes, it includes integration patterns for major platforms like MLflow and Weights & Biases to track training metrics, hyperparameters, and artifacts.

What deployment patterns are covered for production models?

The skill covers several advanced patterns including Canary releases, Blue-Green deployments, and Shadow deployments, along with automated rollback triggers.

MLOps Pipeline Workflow

Name: MLOps Pipeline Workflow
Author: HermeticOrmus

byHermeticOrmus

数据科学与机器学习

Builds production-grade MLOps pipelines by orchestrating data preparation, model training, validation, and automated deployment workflows.

关于

This skill provides a comprehensive framework for managing the complete machine learning lifecycle, from initial data ingestion to production monitoring. It guides users through designing Directed Acyclic Graphs (DAGs) for orchestration, implementing robust data validation, managing experiment tracking, and executing advanced deployment strategies like canary releases. It is particularly useful for engineers and data scientists transitioning from experimental notebooks to scalable, reproducible, and automated production ML systems.

主要功能

0 GitHub stars
End-to-end DAG orchestration patterns for Airflow, Kubeflow, and Dagster
Integration patterns for experiment tracking and model registries
Comprehensive model validation and performance regression detection
Automated data validation and feature engineering pipeline structures
Production deployment strategies including Canary and Blue-Green releases

使用场景

Transitioning local ML experiments into automated production-ready pipelines
Implementing CI/CD for machine learning with automated retraining and deployment
Designing modular and reproducible ML workflows for collaborative data science teams

关于

主要功能

0 GitHub stars
End-to-end DAG orchestration patterns for Airflow, Kubeflow, and Dagster
Integration patterns for experiment tracking and model registries
Comprehensive model validation and performance regression detection
Automated data validation and feature engineering pipeline structures
Production deployment strategies including Canary and Blue-Green releases

使用场景

Transitioning local ML experiments into automated production-ready pipelines
Implementing CI/CD for machine learning with automated retraining and deployment
Designing modular and reproducible ML workflows for collaborative data science teams