How UrbanMove Migrated to Kubernetes and Cut Cloud Infrastructure Costs by 43%
A mobility-tech startup was paying £28,000 per month for infrastructure that could not scale reliably. We migrated them to Kubernetes and rebuilt their cloud foundations while cutting costs by 43%.
Results Achieved
Problem
UrbanMove was running 14 services on a mix of EC2 instances and Elastic Beanstalk environments. Peak traffic regularly caused service degradation because autoscaling was too slow. They were overprovisioning massively during off-peak hours to compensate, resulting in £28,000 monthly AWS bills. They had no container strategy, no infrastructure as code, and their deployment process required an engineer to manually SSH into servers. A planned Series A raised scrutiny on their operational maturity.
Architecture
Cloud Infrastructure & Kubernetes Migration was designed as the operating layer for UrbanMove, integrated into the existing workflow stack and connected to the tools mentioned in the engagement scope. We designed a full cloud-native migration to AWS EKS with Karpenter for node autoprovisioning, spot instance pools for non-critical workloads, and KEDA-based application autoscaling tied to queue depth and HTTP request rate. All infrastructure was rewritten as Terraform modules. Services were containerized with optimized multi-stage Dockerfiles. We implemented Velero for cluster backups, External Secrets Operator for secrets management, and a full observability stack. The migration was zero-downtime using a blue-green DNS cutover strategy.
Solution
We designed a full cloud-native migration to AWS EKS with Karpenter for node autoprovisioning, spot instance pools for non-critical workloads, and KEDA-based application autoscaling tied to queue depth and HTTP request rate. All infrastructure was rewritten as Terraform modules. Services were containerized with optimized multi-stage Dockerfiles. We implemented Velero for cluster backups, External Secrets Operator for secrets management, and a full observability stack. The migration was zero-downtime using a blue-green DNS cutover strategy.
Business Impact
UrbanMove achieved -43% in monthly cloud spend, 3x improvement in peak traffic capacity, -89% in deployment time, New capability in infrastructure provisioning time across a 12 weeks engagement with £58,000 investment.
"The cost savings alone paid for the engagement in month 3. But the real win was passing Series A due diligence — investors flagged our infrastructure as a strength, not a risk."
Cloud Infrastructure Management
Cloud foundations, reliability, security, and cost control for production workloads.
DevOps Automation & CI/CD
Release engineering, deployment automation, platform operations, and observability.
Startup Technical Partner
Fractional CTO leadership and product engineering capacity for growth-stage teams.
The Kubernetes Migration Checklist Nobody Actually Gives You
The real checklist for migrating workloads to Kubernetes — covering the infrastructure, networking, secrets, observability, and the ten things that always break in production.
How I Migrated 2TB of S3 Data and an Entire Kubernetes Cluster Without Losing a Single Night's Sleep
A real-world account of a zero-downtime large-scale cloud migration — covering the data transfer strategy, cluster cutover, DNS flip, and how we validated the entire thing before touching production.
Why Your Startup's Deployment Process Will Break You Before Your Users Do
The hidden technical debt in startup deployment processes — and the four changes that prevent engineering teams from grinding to a halt as the codebase scales.
Ready for similar results?
See how NexForge can design, build, and operate AI-native systems for your business.
Start Your Assessment