Machine learning operations (MLOps)
Azure Machine Learning capabilities that automate and accelerate the machine learning lifecycle
Deliver innovation faster with robust machine learning lifecycle management
MLOps, or DevOps for machine learning, enables data science and IT teams to collaborate and increase the pace of model development and deployment via monitoring, validation, and governance of machine learning models.
Training reproducibility with advanced tracking of datasets, code, experiments, and environments in a rich model registry.
Autoscaling, powerful managed compute, no-code deploy, and tools for easy model training and deployment.
Efficient workflows with scheduling and management capabilities to build and deploy with continuous integration/continuous deployment (CI/CD).
Advanced capabilities to meet governance and control objectives and promote model transparency and fairness.
Build reproducible workflows and models
Reduce variations in model iterations and provide fault tolerance for enterprise-grade scenarios through reproducible training and models. Use dataset and rich model registries to track assets. Enable enhanced traceability with tracking for code, data, and metrics in run history. Build ML pipelines to design, deploy, and manage reproducible model workflows for consistent model delivery.
Easily deploy highly accurate models anywhere
Deploy rapidly with confidence. Use auto-scaling, managed CPU, and GPU clusters with distributed training in the cloud. Package models quickly and ensure high quality at every step using model profiling and validation tools. Use controlled rollout to promote models into production.
Efficiently manage the entire machine learning lifecycle
Use built-in integration with Azure DevOps and GitHub Actions for seamlessly scheduling, managing, and automating workflows. Optimize model training and deployment pipelines, build for CI/CD to facilitate retraining, and easily fit machine learning into your existing release processes. Use advanced data-drift analysis to improve model performance over time.
Achieve governance and control across machine learning assets
Track model version history and lineage for auditability. Use model transparency to understand feature importance and build better models while minimizing bias with fairness metrics. Set compute quotas on resources and apply policies to ensure adherence to security, privacy, and compliance standards. Build audit trails to meet regulatory requirements as you tag machine learning assets and automatically track experiments.
Key phases of MLOps
Build and train reproducible models
Turn your training process into a reproducible pipeline using machine learning pipelines to stitch together all the steps, from data preparation to model evaluation.
Package and deploy models
Package the model into a container image and then deploy it. Use profiling to determine the ideal CPU and memory settings, and to validate models.
Automate workflows, monitor, and manage
Automate the end-to-end machine learning lifecycle with Azure Machine Learning and GitHub to frequently update models, test new models, and continuously roll out new machine learning models alongside your other applications and services.
Apply governance and control
Capture the data required for establishing an end-to-end audit trail of the machine learning lifecycle, including who’s publishing models, why changes are being made, and when models were deployed or used in production.
Resources
Lifecycle infographic
Frequently asked questions about MLOps
-
The rich model registry captures key information such as framework versions, the purpose of the model, model origin, the performance profile, and other data used to operationalize and deploy the model.
-
The audit trail enables automatic tracking of experiments and datasets that correspond to the registered machine learning model and its deployments.
-
The model profiling capabilities help determine optimal execution configurations faster and with higher quality, and provide recommendations for cost versus latency. Model validation enables inferencing against models at scale and validating of output quality.