How To Deploy ML Models With Google Cloud Run

Learn how to deploy Machine Learning models with Google Cloud Run.

#more

Learn how to deploy Machine Learning / Deep Learning models with Google Cloud Run. We build a simple app with TensorFlow and Flask, containerize it with Docker, and deploy it to Google Cloud Run.

You can find the code on GitHub: https://github.com/patrickloeber/ml-deployment.

1. Write App (Flask, TensorFlow)¶

• The code to build, train, and save the model is in the test folder.
• Implement the app in main.py

2. Setup Google Cloud¶

• Create new project
• Activate Cloud Run API and Cloud Build API

3. Install and init Google Cloud SDK¶

• https://cloud.google.com/sdk/docs/install

4. Dockerfile, requirements.txt, .dockerignore¶

• https://cloud.google.com/run/docs/quickstarts/build-and-deploy#containerizing

5. Cloud build & deploy¶

gcloud builds submit --tag gcr.io/<project_id>/<function_name>
gcloud run deploy --image gcr.io/<project_id>/<function_name> --platform managed

Test¶

• Test the code with test/test.py