Deploy Data Science Infrastructure on AWS

This course introduces students to common deployment scenarios for Data Scientists on AWS. It goes over open source tools for deploying infrastructure such as AWS Parallelcluster and Terraform.

Skillset Outcomes

  • Learn to use cookiecutter, a common templating tool that uses the Jinja syntax language.
  • Learn to deploy SLURM Clusters with AWS Parallelcluster.
  • Learn to deploy Kubernetes Clusters with AWS EKS, Rancher, and Terraform.
  • Learn to create reproducible and extendible deployment recipes that can be placed under version control by using Cookiecutter + Terraform. 
  • How to use common CLI programs (Click, Argparse) along with Cookiecutter in order to create recipes that can be used across multiple projects, labs, and institutes.

 

Intended Audience

This course is intended for system administrators or other IT professionals with at least some experience with managing SLURM or Kubernetes clusters. Some programming knowledge is useful, but not necessary.

Course Length

4 days

Course Delivery

This course can be delivered in person or remotely in real-time with Zoom and Slack.

Requirements

Computer

All students require a computer with an internet connection and at least 8 GBs of ram.

IDE

I recommend PyCharm professional, but Visual Studio code can also be used.

Other requirements

All students need Slack for communicating with code blocks, and optionally Zoom if the course will be delivered remotely.

Development Environment

Students can use Docker Desktop on their own laptops, or you can request a development environment for your students, which I build on AWS. Please note that I need 1 week of notice to deploy the development environment, and there is a charge.

Students must also have a way to SSH, either with Putty on Windows or OpenSSH Server on Mac/Linux.

Contact Us

Have question? We'd love to hear from you. Send us a message and we will respond as soon as possible. You can also send an email to [email protected]

Contact Us

Have question? We'd love to hear from you. Send us a message and we will respond as soon as possible. You can also send an email to [email protected]