mlops-pipeline-review
Skill0 starsChanged 5 months agoMITView on GitHub
Review or design an MLOps pipeline — data/feature pipelines, training and experiment tracking, model registry, CI/CD/CT, deployment strategies, monitoring, drift detection, and rollback. Use for ML platform or pipeline work, "how do I productionize this model", or reliability of an existing ML pipeline. For model/system design itself use ml-system-design.
In this repository12 other files it gives its agents.
- adr-writerSkill
- architecture-reviewSkill
- gcp-architectureSkill
- incident-postmortemSkill
- interview-modeSkill
- leadership-lensSkill
- learning-coachSkill
- llm-app-architectureSkill
- ml-system-designSkill
- observability-reviewSkill
- principal-code-reviewSkill
- rag-system-designSkill
The file
Files
MLOps Pipeline Review
Assess the path from experiment to reliable production retraining. Automation and reproducibility are the bar.
Data & features
- Pipeline orchestration & scheduling
- Feature store (offline/online parity)
- Data validation & schema enforcement
Training & tracking
- Reproducible training (pinned data + code + config)
- Experiment tracking & model lineage
- Model registry with stage promotion (staging → prod)
CI/CD/CT
- CI: tests, lint, data validation
- CD: automated deployment with approval gates
- CT (continuous training): retrain triggers (schedule, drift, performance)
Deployment
- Strategy: shadow → canary → progressive rollout
- A/B or interleaving for model comparison
- Rollback plan and versioned artifacts
Monitoring
- Operational: latency, errors, throughput
- Model: input drift, prediction drift, performance decay
- Alerting tied to SLOs
Governance
- Reproducibility & audit trail
- Access control on models/data
- Cost per training run and per prediction
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.