Using an API to Get Predictions

After you deploy a machine learning model on Google Cloud, you can call it from your own application through an API. Below is a simple walkthrough showing how to:

• Set up Google Cloud service account credentials
• Write a Python script to send data to the model
• Create a simple Flask API to return predictions

1. Create a Google Cloud Service Account

First, you need a service account key so your app can access the model.

1. Go to APIs & Services → Credentials
2. Create a service account key
3. Download it as a JSON file

Set this environment variable:

export GOOGLE_APPLICATION_CREDENTIALS="my-service-key.json"

2. Python Client for Sending Data (gcp_client.py)

This script sends data to your deployed model and returns a prediction.

import google.auth
from googleapiclient.discovery import build

credentials, project_id = google.auth.default()
ml = build("ml", "v1", credentials=credentials)

def call_model(features):
    model = "deployed_classifier"
    version = "v1"

    endpoint = f"projects/{project_id}/models/{model}/versions/{version}"
    body = {"instances": [{"x": features}]}

    resp = ml.projects().predict(name=endpoint, body=body).execute()
    return resp.get("predictions", [])

3. Flask App (app.py)

This Flask app exposes a /predict endpoint.

from flask import Flask, request, jsonify
from flask_cors import CORS
from gcp_client import call_model

app = Flask(__name__)
CORS(app)

@app.route("/predict", methods=["POST"])
def predict():
    data = request.get_json(force=True)
    features = data.get("features")

    if features is None:
        return jsonify({"error": "Missing features"}), 400

    preds = call_model(features)
    return jsonify({"prediction": preds[0] if preds else None})

if __name__ == "__main__":
    app.run(debug=True)

4. Run the App

Run this:

export FLASK_APP=app.py
flask run

Test with:

curl -X POST http://127.0.0.1:5000/predict \
  -H "Content-Type: application/json" \
  -d '{"features":[1,2,3,4]}'