This page is also available in Deutsch, Español, Français, Italiano, 日本語, and Português.

K1x Slashes Cloud Costs and Streamlines Kubernetes Operations

How an AI-powered tax software startup right-sized its AKS clusters, shut down over-provisioned node pools, and saved thousands per month

PerfectScale
K1x

The Challenge

After spinning out from an enterprise organization, K1x inherited several Kubernetes clusters but retained none of the team members who originally ran them. Lead platform operator Jason Berk discovered significant over-provisioning: node pools running at two to three times the required minimum scale with less than 10% resource utilization, workloads deployed without best-practice alignment, expensive compute-optimized nodes used unnecessarily, and limited observability—Prometheus surfaced raw metrics but no actionable insights, making optimization 'like trying to find a needle in a haystack.'

The Solution

Rather than spend hours manually figuring out what to measure and how, Jason adopted PerfectScale to deliver immediate, actionable Kubernetes optimization recommendations. PerfectScale provided node-level visibility, identified idle resources, right-sized workload CPU and memory allocations, surfaced memory leaks and other critical issues, and automated tuning—freeing Jason to focus on strategic priorities while managing Kubernetes with a smaller team.

Results

  • Reduced monthly cloud spend by thousands of dollars
  • Shut down entire over-provisioned node pools within a month
  • Right-sized workload CPU and memory allocations without impacting performance
  • Identified a critical issue with one app that was restarting 4,000 times per month
  • Improved developer collaboration with data-backed insights
  • Enabled effective Kubernetes management with a smaller team

PerfectScale gave us the answers we needed on a silver platter. The insights saved us time and money.

Jason Berk, Platform Operator, K1x

K1x Snapshot

K1x is a high-growth, AI-powered tax software provider based in New Jersey that spun out from an enterprise organization to bring its innovative tax solutions to market at scale. Its tech stack includes Azure AKS, ArgoCD, and Terraform. As part of the spinout, K1x migrated several Kubernetes clusters but did not retain the team members who had been running them—leaving lead platform operator Jason Berk to refine the architecture and support a dynamic engineering team.

Identifying Significant Over-Provisioning

When Jason joined K1x, he discovered numerous optimization opportunities across the Kubernetes environment. Node pools were oversized and wasted significant resources, running at minimums two to three times higher than needed while using less than 10% of available capacity. Workloads were deployed without fully aligning with best practices, expensive compute-optimized nodes were used unnecessarily, and existing tools like Prometheus surfaced raw metrics but offered no actionable guidance. As Jason put it: 'We had nodes in our cluster running at a minimum scale that was two or three times what we actually needed. They never auto-scaled up, but we were using less than 10% of the resources and we needed them to scale down. It was a huge waste.'

Replacing Hours of Manual Tuning with Instant Insights

Jason needed a solution that could quickly identify optimization opportunities and provide actionable recommendations. 'I told my CTO: it will take me hours figuring it all out—what to measure, how to measure it, and whether I'm even doing it right—or we can just buy PerfectScale. It'll solve the problem faster and cheaper.' PerfectScale delivered actionable insights ('Here's the workload, here's the right settings'), gave Jason clear visibility into node utilization and workload settings, and simplified Kubernetes optimization in a way that saved him significant time.

Transforming Kubernetes Management

PerfectScale revealed inefficiencies in K1x's Kubernetes environment immediately. 'PerfectScale made me ask the right questions—like, Why are we running this workload on this node pool? Within a couple of months, I shut those node pools down completely and saved us thousands of dollars a month.' The platform helped K1x eliminate idle resources at the node level, right-size workloads with clear CPU and memory recommendations for developers, surface critical issues such as memory leaks (including one app that restarted 4,000 times in a single month before Jason fixed it), and streamline operations through automated tuning.

Monthly Cloud Costs Slashed by Optimizing Node Usage

Within a month, K1x shut down entire node pools, redirecting thousands of dollars in monthly savings back into the budget. Beyond cost savings, PerfectScale eliminated inefficiencies that caused workload instability such as memory leaks and resource bottlenecks, improving resiliency. The tool also enabled Jason to manage Kubernetes effectively with a smaller team.

Why PerfectScale Works for K1x—Beyond Technology

PerfectScale empowered Jason to transform K1x's Kubernetes into a highly optimized environment. With insights, automation, and ongoing support, he can quickly recommend and take action on strategic improvements. Jason also values the partnership: 'I've been blown away by the one-on-one support I've gotten from PerfectScale. They answer my questions quickly and even walk me through problems when I need help. For someone in my position, it's been a lifesaver.'

Learn how PerfectScale improves Kubernetes efficiency

Explore how PerfectScale helps teams right-size clusters, reduce waste, and improve performance without manual tuning.

More customer stories

OneFootball

PerfectScale by DoiT helps OneFootball optimize Kubernetes for global football traffic at scale

25%
reduction in Kubernetes infrastructure costs
80%
reduction in engineering effort spent on Kubernetes cost optimization and resiliency tuning
Luma Health

Luma Health cuts EKS costs by 40% while freeing up thousands of engineering hours a year

90%
Less time spent on manual Kubernetes rightsizing
40%
Reduction in Amazon EKS costs
+1,700h/year
Engineering hours redirected to reliability and performance
PlayHQ

PlayHQ Optimizes Multi-Tenant Kubernetes Costs

40%
Reduction in non-production Kubernetes costs
40%
Reduction in non-production K8s costs
20%
Reduction in production K8s costs
SNCF

How SNCF Cut K8s Waste and Increased Reliability at Scale

30%
More Workloads at Flat Cost
30%
More Workloads Absorbed at Flat Cost
~€500K
Estimated Annualized Savings
NOS

NOS Cuts Kubernetes Costs in Half and Rebuilds Trust in Optimization

50%
Cost Reduction
50%
Cost reduction on largest cluster
0%
Idle resources on main node pools
Trax

How Trax Cut 75% of Kubernetes spend with PerfectScale

75%
Decrease in K8s costs
Riftweaver

How PerfectScale Helps a DevOps Team of One Scale Riftweaver

59%
Reduction in CPU Throttling
6
Critical APIs Improved
80,000+
Game Downloads
Rapyd

How Rapyd Solved Observability Gaps to Cut K8s Costs by 40%

35-40%
Cloud Cost Reduction
35-40%
Projected Cloud Cost Reduction
15+
AWS EKS Clusters Optimized