Machine Learning Ops
MLOps pipelines plus a six-stage recommender framework taken from xAI's open-sourced feed algorithm.
Install
Claude Code: /plugin marketplace add wshobson/agents → /plugin install machine-learning-ops@claude-code-workflows
Model eğitim ve dağıtım hattı kurulurken ya da bir kullanıcı-bağlam çifti için en iyi K öğeyi seçen bir öneri/sıralama sistemi tasarlanırken
- Author
- Seth Hobson
- License
- MIT
The pipeline skill covers the ordinary MLOps spine: DAG orchestration in Airflow, Dagster or Kubeflow, data versioning, experiment tracking, canary and blue-green rollout with a rollback path. The second skill is the unusual one — it encodes the Source, Hydrator, Filter, Scorer, Selector and SideEffect stages from the For You algorithm xAI published, written as an independent reimplementation rather than copied code. Its argument is that most production recommenders are plumbing, not exotic modelling: the scorer changes, the shape of the pipeline does not. That makes the same six stages fit a RAG reranker, a notification triage queue or an alert prioritiser. Three agents split the work between data science, model engineering and the operational side.
Similar skills
Astronomer Agent Skills
SkillAirflow data engineering skills published by the company that maintains Astro.
Data Engineering
SkillFour skills for the batch stack: Airflow DAGs, dbt models, Spark tuning and data-quality contracts.
MLflow Skills
SkillCloses the loop for agent work: instrument, trace, evaluate, change, then verify the change helped.
Comments(0)
Sign in to comment