Authenticate to BigQuery in a Jupyter notebook

If you are working on a cloud-based Jupyter notebook in AI Platform, the instance automatically has access to BigQuery in your GCP project, so you won’t need to set up anything in most cases. Otherwise, run the following steps.

Method 1: with a service account

Create a service account

Follow these steps:

  1. Enable the BigQuery API.
  2. In the GCP console, go to the Create service account key page.
  3. From the Service account list, select New service account.
  4. In the Service account name field, enter a name.
  5. From the Role list, select Project > Owner.
  6. Click Create. A JSON file that contains your key downloads to your computer.

Option A: set the environment variable

Set the environment variable GOOGLE_APPLICATION_CREDENTIALS to the file you downloaded. You can do it in the Terminal, or directly on your Notebook by prefixing ! to the command.

# Set environment variable
!export GOOGLE_APPLICATION_CREDENTIALS="ignore/bigquery_creds.json"
# Start BigQuery client
from import bigquery
bqclient = bigquery.Client()

Option B: specify file location to the BigQuery client

Instead of setting the environment variable, you can directly specify the credentials JSON file when starting BigQuery client, with from_service_account_json().

# Start BigQuery client and specify a credentials file
bqclient = bigquery.Client.from_service_account_json('ignore/bigquery_creds.json')

Method 2: as final user

While the service account method is preferable for most situations, you might need to identify as a final user in some cases, for example to work with datasets that have been granted access to your own account.

Set up

  1. In the GCP console, enable the BigQuery API.
  2. Install the google-cloud-bigquery and google-auth-oauthlib Python libraries:
    pip install --upgrade google-cloud-bigquery google-auth-oauthlib

Create your client credentials

  1. In the GCP console, go to the OAuth consent screen page.
  2. On the Credentials page, click Create credentials, then select OAuth client ID.
  3. Select Other > Create > OK.
  4. Download the credentials by clicking the Download JSON button for the client ID.
  5. Save the file as client_secrets.json

Authenticate and call the API

# Import library and start authentication flow
from google_auth_oauthlib import flow

appflow = flow.InstalledAppFlow.from_client_secrets_file(

# Authentication flow
# If running a remote server, use `appflow.run_local_server()` instead
credentials = appflow.credentials

# Start BigQuery client
bqclient = bigquery.Client(project='my-project-id', credentials=credentials)