Start here in MLOps
The most useful guides in this topic, not just the most recent.
MLOps Guides & Tutorials
Operationalise machine learning — MLflow, Docker, Kubernetes, CI/CD, monitoring and production ML system design.
MLOps articles
Prometheus and Grafana for ML Monitoring
Prometheus and Grafana are the usual open-source pair for watching services. This guide instruments a real model API, scrapes it with Prometheus 3.15, unit-tests a drift alert, provisions a Grafana 13 dashboard from files and replays two real weeks of requests to see what the setup catches and what it misses.
Real-Time Feature Engineering for Online Models
Online models need features from events seconds old, and those features must match the ones the model was trained on. On a month of real New York taxi pickups, this guide builds a streaming window counter, shows a one-character boundary change breaking 3.7% of lookups, and measures what a two-minute lag costs.
Secrets Management for ML Pipelines
ML pipelines handle more credentials than most code, and leak them in places ordinary software does not, such as experiment trackers. This guide plants AWS's published example key in a project and follows it: a scanner flags it, MLflow stores it in plain text, a log filter redacts it and a vault keeps it out.
Shadow Deployment and Canary Releases for ML
Shadow mode runs a new model on real traffic without letting it answer; a canary lets it answer for a few users first. On real e-commerce sessions, this guide shows a unit bug that 89% agreement hid, the flag-rate guardrail that caught it, and why per-stage business metrics were too noisy to gate on.
Model Explainability in Production with SHAP
SHAP turns a prediction into per-feature contributions you can show a customer or an auditor. On 30,000 real credit card accounts, this guide produces reason codes for declined applicants, checks that explanations add up exactly, and shows an equally valid explainer setting changing the reasons for 42% of them.
Airflow for ML Pipelines: A Practical Guide
Airflow can schedule more than data loads: it can retrain a model, score each day, check the error and raise an alert. This guide builds a training DAG and a scoring DAG on real bike rental data, and shows them handling a normal day, a bad day, and the day Hurricane Sandy cut the data off.