ML model deployed on GCP via Flask

In this project, an ML-model is deployed on GCP using Flask. The Machine Learning step of this project is coming soon.

Here is the roadmap to build a continuous delivery of an application through Google Cloud Platform.

Step-by-step roadmap

1. The files are created on GitHub. I first create a project on GCP called Continuous-Delivery-ML-model.

2. Let us make a new directory MLdelivery

mkdir MLdelivery

cd MLdelivery/

cd GCP-ML-Model-via-Flask

3. Now I clone the files on my GitHub repository "GCP-ML-Model-via-Flask"

4. We need a virtual environment:

virtualenv ~/.cd

source ~/.cd/bin/activate

5. Install the dependencies and run the application. We see the link to open up our application on the web browser.

A message Running on http://127.0.0.1:8080 (Press CTRL+C to quit) will appear.

make install

python main.py

---

Until now we have built a Flask application which works well; yet we haven't built the continuous delivery.

6. Let us deploy it. Go to the main terminal and put

gcloud app deploy

Gcloud will be requesting the credentials to make a GCP API call. Authorize Cloud Shell.

Choose the region.

7. Good! The application is online. Here it is...

https://continuous-delivery-ml-model.ew.r.appspot.com/

and this...

https://continuous-delivery-ml-model.ew.r.appspot.com/echo/firat

8. Go to Cloud Build on GCP in order to create a new trigger cloud-delivery

Select the options below:

• Repository event that invokes trigger: Push to a branch
• Branch: ^master$

Connect New Repository. Here we need to autorize API to connect to our GitHub repo.

We now make some settings. Enable App Engine Admin API and Service Accounts

9. Any event on the code on the GitHub repo will directly give the result on the web app. Make a change and see it.

**UPDATE : ** The web application is no more working since I have run out of my free GCP account.