Skip to content

Latest commit

 

History

History
99 lines (80 loc) · 3.8 KB

README.md

File metadata and controls

99 lines (80 loc) · 3.8 KB
Raw

MLOps Hackathon

Learn about MLOps by deploying your own ML pipelines in Google Cloud. You'll solve a number of exercises and challenges to run pipelines in Vertex AI, continuously monitor your models, and promote your artifacts to a production environment.

Getting started

As a hackathon attendee, simply follow this notebook series in your Vertex AI Workbench instance:

  1. Health check - start here
  2. Run pipelines
  3. Promote model
  4. Challenge: Model monitoring
  5. Challenge: Real-time predictions

❗Note: This workshop has been designed to be run in Vertex AI Workbench. Support for running the workshop locally is provided, but we recommend Vertex AI Workbench for the best experience.

For instructors

Shell

The notebooks are self-contained but instructors of this hackathon are asked to prepare the following for hackathon attendees.

Deploy projects

  1. Create 3x Google Cloud projects (dev, test, prod)
  2. Use make deploy to deploy resources in each of them. Checkout for the workflow infrastructure setup notebook. 
    # ensure env.sh references dev project
make deploy env=dev
# ensure env.sh references test project
make deploy env=test
# ensure env.sh references prod project
make deploy env=prod

Configure dev project

  1. Update terraform/users.tfvars
  2. Ensure env.sh references the dev project and run make deploy env=dev

Configure test project

  1. Add each user with their Google account with the following Viewer role

  2. Create Cloud Build repo: Test repo Note: You might need to enable the Secret Manager API.

  3. Create Cloud Build Service Account with IAM roles:

    • Artifact Registry Writer
    • Logs Writer
    • Storage Admin
    • Vertex AI User
    • Grant role "Service Account User" for this SA on [email protected]

  4. Create 3x Cloud Build triggers. For each use the settings below and reference the corresponding cloudbuild/<trigger>.yaml file: Test triggers

    • Event: Pull Request
    • Source:
      • Repository: 2nd gen
      • Base branch: ^develop$
    • Comment control: Required
    • Configuration:
      • Type: Cloud Build configuration file
      • Location: Repository
      • Cloud Build configuration file: /cloudbuild/<trigger-name>.yaml
    • Advanced:

    Note: Trigger e2e-tests requires the following substitution variables: Test repo

Configure prod project

  1. Add each user with their Google account with the following Viewer role
  2. Create Cloud Build repo: Test repo
  3. Create 1x Cloud Build trigger:
    • Event: Push new tag
    • Source:
      • Repository: 2nd gen
      • Base branch: .*
    • Configuration:
      • Type: Cloud Build configuration file
      • Location: Repository
      • Cloud Build configuration file: /cloudbuild/release.yaml
    • Advanced:
  4. Create Cloud Build Service Account with IAM roles:
    • Artifact Registry Writer
    • Logs Writer
    • Storage Admin
    • Vertex AI User

Clean up

  1. Shutdown Workbench instances at the end of the day
  2. Delete user branches in this repo
  3. Delete user tags in this repo
  4. Delete 3x projects