Automation and scaling

MLOps Concepts

Folkert Stijnman

ML Engineer

Automation and scaling

end-to-end pipeline

MLOps Concepts

Automation and scaling

Automation and scalability

MLOps Concepts

Design phase

Design phase components

Project design

  • Project design remains a manual process
  • Use templates to automate and scale
MLOps Concepts

Design phase

Design phase project design

Project design

  • Project design remains a manual process
  • Use templates to automate and scale

Data acquisition

  • Can be automated
  • Enables high data quality
MLOps Concepts

Development phase

Development phase feature store

Feature store

  • Saves time building the same features
  • Helps to scale
MLOps Concepts

Development phase

Development phase experiment tracking

Feature store

  • Saves time building the same features
  • Helps to scale

Experiment tracking

  • Automates tracking
  • Ensures reproducibility
MLOps Concepts

Deployment phase

Deployment phase containerization

Containerization

  • Easy to start up copies of the same application
  • Improves scalability
MLOps Concepts

Deployment phase

Deployment phase ci/cd pipeline

Containerization

  • Easy to start up copies of the same application
  • Improves scalability

CI/CD pipeline

  • Automates development and deployment
  • Increases velocity of processes
MLOps Concepts

Deployment phase

Deployment phase microservice architecture

Containerization

  • Easy to start up copies of the same application
  • Improves scalability

CI/CD pipeline

  • Automates development and deployment
  • Increases velocity of processes

Microservices architecture

  • Improves scalability
  • Independent development and deployment
MLOps Concepts

Let's practice!

MLOps Concepts

Preparing Video For Download...