A Cloud-Powered Boom

Imagine you’re leading engineering at a fast-growing tech startup. Your team has adopted Kubernetes on AWS Elastic Kubernetes Service (EKS) to scale applications with ease.

The cloud powers rapid feature releases and meets customer demand effortlessly. Infrastructure scales like magic.

Then the AWS bill arrives. What was once a reasonable cost has spiraled into a daunting figure.

You’re not alone. at nOps, we saw this story play out across countless organizations adopting Kubernetes. The promise of scalability was real, but so was the challenge of runaway costs.

Kubernetes: Power and Complexity

Kubernetes has transformed application deployment. AWS EKS is a go-to for teams leveraging this power in the cloud.

But with growth comes complexity. Compute costs often dominate AWS bills—sometimes 70% or more.
Engineering teams shared their struggles with us. Managing costs felt overwhelming without constant oversight.

Resources were overprovisioned. Savings Plans went unused. Spot instances, though cost-effective, required too much manual effort.

The Process: From Chaos to Control

Here’s how Compute Copilot EKS transforms the story, step by step:

Monitor in Real-Time: Dashboards I designed give you a granular view of Container Efficiency, Node Efficiency and Price Efficiency. All of them updating as Compute Copilot adjusts on the fly.

Quick Integration: Plug it into your AWS EKS environment in minutes. It syncs with Cluster Autoscaler or Karpenter, respecting your existing setup.

Analyze and Optimize: Our ML models kick in, scanning usage patterns to rightsize containers and place nodes with precision.

Automate Savings: It provisions the cheapest, most stable compute—maximizing Spot usage with termination protection and ensuring 100% commitment utilization.

Results That Speak

This is one of the most impactful Compute CoPilot features that I led design for, and it is being used now by +600 businesses Saving them millions of dollars with a percentage of 60% - 70% Savings

The Process Behind the scenes 🤔

Discovery Phase

Stakeholder Interview

 I started with interviewing internal stakeholders, especially the Devops team who are the main persona for this feature and that helped us a lot in shaping the Compute CoPilot feature sharp enough to the point where it provides huge value and savings for the customers who use Kubernetes.

Outcomes of the interviews

• Dashboard that provides full visibility and monitoring for what’s happening under the hood of these services and also of nOps automation solution

• Basic and important Insights into Kubernetes Clusters that would be an alternative for DevOps Engineers instead of using Lens IDE.

• The need to have an automated solution for managing the optimization of EKS Clusters. Rightsizing Containers and Clusters Spot Coverage

• Services Cost and how much nOps can save customers and show realization after configuration 

• Simple and intuitive configuration process

• Flexible configuration for different environments

• Configuration of different types of Kubernetes Clusters

Competitors Analysis

The main goal was to understand the various approaches and patterns used for managing Kubernetes and how they provide automation solutions and visibility

Key Takeaway/Observations from the Competitors Analysis:

• Most of the competitors are only focused on containers rightsizing optimization

• The navigation is somehow time consuming and challenging to take actions on optimization and savings features as you have to go through different features and enable them one by one for each cluster

• Most of the competitors don’t have a clear interface to configure spot management

• Couple of the competitors provide an Audit log to track events happening in the background

• Most of their navigation is basically around Cluster where the main navigation item is the cluster and everything else is under it

Define Phase

Affinity mapping

I worked with the team to brainstorm ideas and provide a solution for the problem defined. We used Miro sticky notes, and we took some time to go through the activity. Then we categorized the notes, voted on the ideas that we agreed on, and sorted out the priorities.

User Flow

Before jumping into sketching or wireframing I had to map out the user flows to get the full picture and process clear to everyone involved on the feature. Where does the flow start? What are the actions? How to land on these actions? Etc..

Initial soft mockups

After getting a clear picture of what we have in mind I started by sketching the ideas then by the help of our design system components library I created a soft mockup for the EKS Dashboard and the configuration process.

Internal Design critique

The internal design review call was so productive, I reviewed the designs with the Devops team and engineers to answer a couple of questions

• Is the information we show valuable? Do you see everything you need to take action?

• Is the configuration process clear and intuitive?

• As a DevOps or an engineer Would you use this?

We had valuable feedback to iterate on the designs to adjust some of the information we show, provide table filtering, Date filtering and other updates

The Final Frame: Your Next Chapter
If you’re an engineering leader wrestling with EKS costs, Compute Copilot EKS is your plot twist. As the designer behind it, I’m proud to have built something that delivers not just savings, but freedom—freedom to innovate without the weight of runaway bills.

Ready to see it in action? Check out the feature page