skills.rest
MonumentalSystems
MonumentalSystemsOFFICIAL
Explore all skills by this author →
WebsiteGitHub

mlops-pipelines

Design MLOps pipelines for batch, real-time, edge, and streaming model deployment.

17|5|Updated Feb 6, 2026
Search Tags:#ci/cd#monitoring#drift detection#mlops#model deployment#model registry#feature store
One-click install
npx skills add https://github.com/MonumentalSystems/Atlas-Agent-Teams --skill mlops-pipelines-monumentalsystems
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: mlops-pipelines
Source: https://github.com/MonumentalSystems/Atlas-Agent-Teams/tree/main/teams/data-science/skills/mlops-pipelines
Command: npx skills add https://github.com/MonumentalSystems/Atlas-Agent-Teams --skill mlops-pipelines-monumentalsystems

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps you operationalize machine learning by turning trained models into reliable production services with repeatable deployment, continuous monitoring, and drift detection.

Core Features & Use Cases

  • Model deployment strategies: Covers batch, real-time API serving, edge deployment, and streaming inference patterns for different latency and scale requirements.
  • Monitoring and drift detection: Defines key performance metrics, data/label drift types, and mitigation approaches with alerting and incident response.
  • CI/CD for ML: Provides an end-to-end pipeline view (ingestion → feature engineering → training → evaluation → deployment → monitoring), including canary/A-B testing and rollback.
  • Feature store concepts: Explains offline/online store design, feature registry governance, and versioned feature lineage.
  • Model versioning and registry: Describes metadata, artifacts, lifecycle stages, and promotion controls to improve reproducibility and governance.

Quick Start

Ask for an end-to-end MLOps plan for a model that needs real-time predictions, monitoring, drift alerts, and a safe rollout strategy.

Frequently Asked Questions about mlops-pipelines

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

FAQPage Schema
How do I set up an MLOps pipeline for real-time model deployment?▼

To set up an MLOps pipeline for real-time model deployment, you design inference patterns for API serving alongside continuous monitoring, drift detection, and CI/CD rollback controls.

What is data drift detection in machine learning?▼

Data drift detection in machine learning identifies shifts in input data distributions or labels over time. It triggers mitigation approaches, alerting, and incident response to maintain model reliability.

How do I use a feature store for model versioning?▼

Using a feature store for model versioning involves governing an offline and online store registry with versioned feature lineage, ensuring reproducible artifacts and lifecycle promotion controls.

Does CI/CD for ML include canary testing and rollback?▼

Yes, CI/CD for ML includes canary and A/B testing with rollback capabilities. The pipeline spans ingestion, feature engineering, training, evaluation, deployment, and monitoring.

What's the best way to monitor ML models in production?▼

The best way to monitor ML models in production is defining key performance metrics and drift categories to trigger retraining and mitigation, ensuring dependable batch, streaming, or edge usage.

Related Skills

Explore More Data & Analytics Skills→
tonone-aitonone-ai
69

cortex-model

Automate end-to-end ML pipelines from data ingestion to serving endpoints.

Official
Advanced
Ethical-AI-SyndicateEthical-AI-Syndicate
1

model-lifecycle-manager

Manage AI/ML model registration, versioning, monitoring, drift detection, and retraining triggers.

Official
Advanced
KentoShimizuKentoShimizu
7

ml-feature-engineering

Catalog ML feature engineering workflows for feature definition, lineage, and train-serve parity.

Community
Advanced
grasberggrasberg
3

ml-ops-engineer

Coordinate ML model lifecycles from notebook to production deployment.

Community
Advanced