top of page

Supercharged Marketing with Kubernetes


Technologies: Python, k8s, Docker, Helm

The customer wanted to supercharge their 12-person marketing department but was stuck on legacy software. The project had two main goals: upgrade to modern software and improve access to computing power. I configured JupyterHub to run inside a Kubernetes cluster to do both.


I learned Kubernetes with dockerized containers and deployed Jupyterhub for a twelve with a built-in data science environment but we had one major problem: pods inside the cluster were being orphaned at random, sometimes while they were spawning. The main purpose of a Kubernetes cluster is for it to make sure the pods described inside are healthy and running when they should be, if it "forgets" about them that means the Kubernetes engine is failing its primary purpose.


After an investigation and several covert meetings with the engineer who designed the engine for this cloud provider, I decided to patch it myself by modifying the scheduler configuration to spam all the pods with a ping every second. This ensured that the service operated smoothly.


The configuration worked and provided access to the customer with the tools they needed to expand their business and make sales.

Do you have an ambitious project? Email me and let us work together:

bottom of page