brightsparc / amazon-sagemaker-kernels

Bring your own SageMaker Studio Image for running R notebooks.

Bring your own SageMaker Studio Image

This repository provides you a quick start for Bringing your own custom container image to Amazon SageMaker Studio notebooks.

Prerequisites

In order to build a custom kernel your Amazon SageMaker Studio Execution role will require additional permissions, see the PRE-REQS.md for applying to an existing role.

Alternatively, you use the AWS CloudFormation stack cloudformation\sagemaker-studio-infra.yaml to setup a new SageMaker Studio domain and Execution role with sufficient permissions, or click the following button to launch the stack in us-west-1 region.

This stack will return output parameters for the following resources

1. SageMaker Default S3 Bucket
2. SageMaker Studio Domain ID
3. SageMaker Studio Execution Role

Open SageMaker Studio and Clone the notebook

1. Open the SageMaker console
2. Chose Amazon SageMaker Studio at the top left of the page.
3. On the Amazon SageMaker Studio Control Panel, choose the default username and then choose Open Studio.
4. Follow the instructions to Clone a Git Repository and provide the repository https://github.com/brightsparc/amazon-sagemaker-kernels

Open the cloned folder to browse to the r-kernel folder.

Build R Notebook Kernel for Sagemaker Studio

In the r-kernel example we use the BUILD.ipynb notebook to demonstrate how to use the Sagemaker Studio Image Build CLI to build and register a customer Notebook Kernel from within SageMaker Studio.

This allows allows you to run R code inside a SageMaker Studio Notebook.