Amazon Sagemaker is a fully managed service for handling machine learning workflows. It enables users to build, train, and deploy ML models quickly. But before you can deploy your ML model, it must first be built, tuned, and iterated. Notebooks, and more specifically Zepl notebooks, are perfectly suited for these tasks.
Now you can use Zepl to connect directly to Sagemaker in your VPC by simply selecting the Sagemaker resource available in Zepl.
Simply create a new notebook and select Sagemaker in the resource drop down:
You can also switch an existing notebook's resource to Sagemaker by clicking the Settings link in the top right of the notebook.
That's it! Nothing else to install. You can then do the following in the notebook:
%python from sagemaker.session import Session from sagemaker import KMeans import boto3 import pickle, gzip, numpy, urllib.request import matplotlib.pyplot as plt AWS_ACCESS_KEY_ID="[your_AWS_KEY_ID]" AWS_SECRET_ACCESS_KEY="[your_AWS_SECRET_KEY]" REGION_NAME = "[region_of_your_VPC]" def get_boto3_session_with_credentials(): return boto3.Session(aws_access_key_id=AWS_ACCESS_KEY_ID, aws_secret_access_key=AWS_SECRET_ACCESS_KEY,region_name=REGION_NAME) role = "[your_ARN_ROLE]" session = Session(get_boto3_session_with_credentials()) bucket = session.default_bucket() # your code goes below
As long as you have your AWS credentials setup correctly your model would be deployed to the Sagemaker service in your VPC. In addition, the Zepl Sagemaker resource image is already pre-loaded with the following Python libraries allowing you and your team to leverage Sagemaker for all of your machine learning needs: