Introduction

"Silver Bullet on Deploying Any Data Science Projects to IBM Cloud Pak for Data" aims to deploy any python/R Data Science projects and ML models to IBM Cloud Pak for Data (CPD) on-prem or CPDaas. With a simplified and consistent interface, users without any prior knowledge of IBM Cloud Pak for Data can be enabled in several minutes.
By configuring a self-explained YAML file, all the deployments can be done with one-line script, which can be used to enable CI/CD very easily.

Features

• Support projects developed within or outside of Watson Studio in IBM Cloud Pak for Data.
• Support CPD 3.5/4.0/4.5/4.6/4.7/4.8 and potentially all future versions
• Support Python/R Scripts, Python/R notebooks, R Shiny.
• Support "batch job" and "online" deployment
• Support all ML frameworks, as long as it can be done with python/R.
• Email notification can be enabled if there is an accessible SMTP server. It is particularly useful when python/R failed due to bugs or imperfect data.
• Contain stderr and stdout in job logs and easy to know what happened in detail.

Authors: Jianbin Tang, [email protected]; Rutvik Dave, [email protected]

Also appreciate help from:

• Xavier Mary
• Cedric Jouan
• Randy Phoa
• Carlos Mejia Johnson

Usage

There are 2 ways to deploy.

• With cpdctl, please refer to deployment_with_cpdctl
• With wml client, please refer to deployment_with_wml_client

Code Structure

• deployment_with_cpdctl :

  • Support py script and notebook deployment.
  • Support R shiny deployment.
  • Support deploy to SPARK and any customised environment.
  • Do not support py or R script as "online" API yet.

• deployment_with_wml_client

  • Support py script deployment as job or online API
  • Support r script deployment as online API
  • Do not support Jupyter notebook directly and only can be deployed after converting it to py.
  • Do not support deploy to SPARK environment and any customised environment.

• py_examples : contains examples for python

• r_example : contains examples for R

License

• Apache License 2.0.
• Feel free to fork it!
• Not a license requirement, but if you like it or used it, appreciate your "watch" and "star" the project :)
• Welcome your feedbacks and contributions.

Thank you!