On this page
The Challenge
- Scaling by hand. Manual scaling could not keep up with weekend traffic surges or month-end ETL workloads.
- Slow, error-prone releases. Release cycles caused delays and strained the team.
- Inconsistent environments. Complex resource management and inconsistent environments hurt productivity and performance.
What SquareOps Did
SquareOps moved the workloads onto Amazon EKS with Karpenter node autoscaling and introduced Atmosly as the self-service platform for deployments and environments.
- Autoscaling on EKS. Karpenter adds and removes capacity as demand changes, so weekend traffic and month-end ETL runs are absorbed without anyone resizing the cluster.
- Built-in CI/CD. Atmosly pipelines automated updates and rollbacks, which reduced downtime, improved stability and sped up feature delivery.
- Consistent environments. Environments are created and scheduled through the platform instead of by hand, including scaling them up or down and starting and stopping them as needed.
- Right-sized capacity. SquareOps also moved suitable workloads onto cost-efficient capacity such as spot instances and stopped non-production resources off-peak.
- Quick adoption. The platform was simple enough for the tech team to adopt without extensive training.
Results
- Automatic scaling. Spikes and heavy ETL loads are handled without manual intervention.
- Faster, safer releases. Release cycles became quicker and less error-prone.
- More time for the product. Automating repetitive tasks freed the tech team to build features instead of maintaining infrastructure.
Conclusion
With EKS, Karpenter and an automated delivery platform underneath, 32nd Avenue turned manual scaling and slow releases into a setup that absorbs peak demand and ships changes faster, and that is ready to grow with the business.

