claudegoodies
Skill

mle-workflow

From affaan-m

Production machine-learning engineering workflow for data contracts, reproducible training, model evaluation, deployment, monitoring, and rollback. Use when building, reviewing, or hardening ML systems beyond one-off notebooks.

Guides ML systems through data contracts, reproducible training, evaluation gates, deployment, monitoring, and rollback.

Use it when

  • Converting notebook code into a training/eval/inference pipeline
  • Designing model promotion criteria, offline/online evals, or rollback paths
  • Debugging data drift, label leakage, stale features, or train/serve mismatch
  • Adding monitoring, canary rollout, or shadow traffic for a deployed model

Skip it if

  • One-off notebook experimentation with no production deployment planned
  • Relies on many other named skills (python-testing, deployment-patterns, etc.) for actual implementation
  • No concrete instructions here, mostly a routing table pointing to other skills/agents

Facts

Repository
affaan-m/ECC
Status
Actively maintained
Last commit
Declared tools
Read, Write, Edit, Bash, Grep, Glob
Source file
.agents/skills/mle-workflow/SKILL.md

Source preview

The instructions Claude Code reads when this skill runs.

# Machine Learning Engineering Workflow

Use this skill to turn model work into a production ML system with clear data contracts, repeatable training, measurable quality gates, deployable artifacts, and operational monitoring.

## When to Activate

- Planning or reviewing a production ML feature, model refresh, ranking system, recommender, classifier, embedding workflow, or forecasting pipeline
- Converting notebook code into a reusable training, evaluation, batch inference, or online inference pipeline
- Designing model promotion criteria, offline/online evals, experiment tracking, or rollback paths
- Debugging failures caused by data drift, label leakage, stale features, artifact mismatch, or inconsistent training and serving logic
- Adding model monitoring, canary rollout, shadow traffic, or post-deploy quality checks

## Scope Calibration

Use only the lanes that fit the system in front of you. This skill is useful for ranking, search, recommendations, classifiers, forecasting, embeddings, LLM workflows, anomaly detection, and batch analytics, but it should not force one architecture onto all of them.

- Do not assume every model has supervised labels, online serving, a feature store, PyTorch, GPUs, human review, A/B tests, or real-time feedback.
- Do not add heavyweight MLOps machinery when a data contract, baseline, eval script, and rollback note would make the change rev
View full source on GitHub →

Other skills