ml-ops-engineer — Claude Code skill: model deployment and monitoring
ml-ops-engineer is a skill for MLOps automation: deploying models, orchestrating training pipelines, monitoring feature drift, and configuring feature stores. It addresses gaps in ML deployment workflows.
LicenseRef-MIT-Commons-Clause 180 installs ★ 1,183 updated 2026-10-23
npx skills add borghei/Claude-Skills --skill ml-ops-engineer -a claude-codeWhat ml-ops-engineer does
This skill structures the full MLOps lifecycle: assessing maturity (manual notebooks → automated pipelines → full CI/CD), deploying models, registering artifacts, and setting up monitoring. It asks clarifying questions upfront about serving mode (real-time FastAPI or batch), latency SLA, current automation level, and drift thresholds. These drive the architecture choice.
Includes ready-to-use examples: FastAPI server for real-time predictions with latency tracking, Kubernetes YAML with autoscaling, Python functions for drift detection using Kolmogorov–Smirnov tests, alert rule templates for latency and error rate, and a Feast feature store scaffold. The workflow covers training pipeline construction with accuracy gates, MLflow artifact logging, health checks for serving, smoke tests, and rollback verification.
Once it understands your constraints, it identifies the highest-impact maturity gap to close first and provides concrete steps to close it.
What it helps with
- deploy a trained model to production with real-time latency SLA
- set up automatic retraining when feature drift is detected
- log model versions, metrics and artifacts to a central registry
- monitor P99 latency, error rates and prediction distribution shifts
- build CI/CD for ML pipelines from data ingestion to serving
Example prompts for the agent
- Deploy fraud-detector to Kubernetes with 3 replicas and CPU-based autoscaling
- Set up daily drift monitoring for the feature store with alerting thresholds
- What steps do I need to move from manual notebooks to an automated training pipeline?
How to install ml-ops-engineer
1Skills CLI (Vercel)
npx skills add borghei/Claude-Skills --skill ml-ops-engineer -a claude-code2Manually: put the files in place
~/.claude/skills/ml-ops-engineer/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/ml-ops-engineer/
1Skills CLI (Vercel)
npx skills add borghei/Claude-Skills --skill ml-ops-engineer -a codex2Manually: put the files in place
.agents/skills/ml-ops-engineer/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add borghei/Claude-Skills --skill ml-ops-engineer -a cursor2Manually: put the files in place
.cursor/skills/ml-ops-engineer/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install borghei/Claude-Skills data-analytics/ml-ops-engineer/SKILL.md2Manually: put the files in place
.github/skills/ml-ops-engineer/SKILL.md1Skills CLI (Vercel)
npx skills add borghei/Claude-Skills --skill ml-ops-engineer -a gemini-cli2Platform CLI
gemini skills install https://github.com/borghei/Claude-Skills --path data-analytics/ml-ops-engineerManually: put the files in place
.devin/skills/ml-ops-engineer/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add borghei/Claude-Skills --skill ml-ops-engineer -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about ml-ops-engineer
How do I install ml-ops-engineer in Claude Code?
Run npx skills add borghei/Claude-Skills --skill ml-ops-engineer -a claude-code, then check the docs at ~/.claude/skills/ml-ops-engineer/SKILL.md.
What does ml-ops-engineer do?
It helps you deploy models to production, set up training pipelines, enable drift monitoring, and automate ML CI/CD. It asks upfront about serving mode, latency SLA, and current automation level to shape the architecture.
What tools does it support?
It works with FastAPI for serving, Kubernetes for orchestration, MLflow for model registries, Feast for feature stores, Airflow or Kubeflow for pipelines, and standard monitoring tools like Prometheus.