skills.rest
kalilurrahman
kalilurrahmanCOMMUNITY
Explore all skills by this author →
Website𝕏@krahmanGitHub

model-versioning

Implement ML model registry management with lineage tracking, artifact storage, and stage transitions.

1|Updated Mar 21, 2026
Search Tags:#rollback#mlops#lineage tracking#model deployment#mlflow#model registry#model versioning
One-click install
npx skills add https://github.com/kalilurrahman/kr-claudiator-skills-original-prompts --skill model-versioning-kalilurrahman
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: model-versioning
Source: https://github.com/kalilurrahman/kr-claudiator-skills-original-prompts/tree/main/04-ai-ml/model-versioning
Command: npx skills add https://github.com/kalilurrahman/kr-claudiator-skills-original-prompts --skill model-versioning-kalilurrahman

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires mlflow.

What problem does it solve? ML teams lose track of which model version serves production, what data trained it, and who approved its promotion, making rollbacks and audits impossible. This Skill establishes a complete model versioning workflow with registry management, lineage capture, and controlled stage transitions. ## Core Features & Use Cases - MLflow Registry Management: Register every trained model with automatic lineage tags including run ID, git commit, data hash, and framework. - Audited Stage Transitions: Promote models through None, Staging, Production, and Archived stages with mandatory approval comments and actor tracking. - Rollback and Comparison: Roll back to archived versions with full audit trails and compare challenger versus champion metrics side by side. - Use Case: A data science team promotes a new churn model to production via a GitHub Actions workflow that requires explicit approval, automatically archives the previous version, and records who approved the transition and why. ## Quick Start Set up an MLflow model registry with lineage tagging and a staging-to-production promotion workflow for my trained model.

Frequently Asked Questions about model-versioning

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I register a model in the MLflow Model Registry?▼

Call mlflow.register_model with a runs:/ URI pointing to the run's artifact path and a model name. Then use MlflowClient to set descriptions and lineage tags like run_id, git_commit, and data_hash on the registered version.

How to promote an MLflow model from Staging to Production?▼

Use transition_model_version_stage on MlflowClient with the target stage. Production promotion should require an explicit approval comment and approver, and existing Production versions should be archived first to maintain a clean lifecycle.

What metadata should I track for ML model lineage?▼

Track the training run ID, git commit SHA, training data hash, registering user or service account, and framework type as model version tags. This enables full traceability from a served model back to its code and data.

How do I roll back a production ML model to a previous version?▼

Query the registry for the most recent Archived version, archive the current Production version, then transition the archived version back to Production. Record the rollback reason and approver in audit tags for compliance.

Should I delete old model versions from the registry?▼

No, archive them instead of deleting. Deleted versions break rollback capability and lineage history, while archived versions remain queryable and restorable when a production issue requires reverting.

Related Skills

Explore More Data & Analytics Skills→
box755box755

mlflow

Track ML experiments and manage model artifacts with MLflow.

Community
Intermediate
Nir-BhayNir-Bhay
19

ml-api-endpoint

Deploy machine learning models as FastAPI inference endpoints with Dockerfile support.

Community
Advanced
KentoShimizuKentoShimizu
7

mlops-monitoring-drift

Detect data drift, concept drift, and quality degradation in production ML pipelines.

Community
Advanced
miyake-sanmiyake-san

mlops-engineer

Designs ML pipelines, experiment tracking, and model registries with MLflow, Kubeflow, and cloud platforms.

Community
Intermediate