Orchestration in MLOps

Fully Automated MLOps

Arturo Opsetmoen Amador

Senior Consultant - Machine Learning

A core component in MLOps systems

Image of the MLOps reference architecture with two elements highlighted: orchestrated experiments in the development environment and the automated pipeline in production.

Modularity & reusability

Figure with two ml pipelines, an upper and a lower one. The top pipeline is in the development and experimentation environment and the bottom one in production.

Orchestration & automation

ML pipelines are used in:

Development & experimentation

Figure of a component: orchestrated experiment.

Production

Figure of an automated ML pipeline.

Direct Acyclic Graphs in MLOps

Graphical pipeline representations
Pipeline steps represented as nodes
Steps dependencies are edges

Image of a simple DAG with interconnected steps in a ML pipeline.

What is orchestration in MLOps?

Managing & automating task flows

Scheduling, monitoring of tasks

Managing data dependencies and flows

ML pipelines - development & experimentation

Figure of an orchestrated ML pipeline in the development & experimentation environment. The pipeline is triggered by a data scientists and orchestrated by an orchestrator component.

Manage end-to-end tasks in model training
Correct flow, record and logging
Parallel experimentation

ML pipelines - production

Figure of an automated ML pipeline in the production environment. The pipeline is triggered by an automated trigger component and orchestrated by an orchestrator component.

Helps automating model deployment
Manages and execute pipeline steps
Consistent and reliable deployments

Let's practice!

Fully Automated MLOps