Cloud Cost Optimization: How a Last-Mile Delivery Startup Cut AWS Spend by 40% with Karpenter and Spot Instances
Comprehensive FinOps engagement reducing AWS costs 40% for delivery startup using Karpenter auto-scaling, 70% Spot instance coverage, and API Gateway optimization in 6 weeks.
Key Results
Monthly AWS spend reduced significantly
Node provisioning from 10+ minutes to under 2 minutes
Workloads running on Spot instances
From audit to full implementation
What We Were Solving
A Series A funded last-mile delivery startup operating across six major US markets was struggling with rapidly escalating AWS costs. Their delivery volume was growing, but cloud spend was growing even faster — threatening their runway and unit economics.
The challenges:
- AWS costs growing faster than revenue, squeezing margins on $4-6 deliveries
- Over-provisioned Kubernetes nodes running 24/7 regardless of delivery volume
- API Gateway costs spiking with high request volumes from carrier and merchant integrations
- Slow scaling during peak delivery windows causing driver app latency
- No visibility into cost allocation across their five microservices
How We Solved It
We conducted a comprehensive FinOps review of their AWS infrastructure and implemented optimizations across compute, API management, and operations — without sacrificing performance or reliability.
Compute Optimization with Karpenter
Replaced the static node groups with Karpenter for intelligent, just-in-time node provisioning. Configured Spot instance pools with automatic fallback to on-demand, achieving 70% Spot coverage during normal operations. Nodes now scale in minutes instead of waiting for scheduled scaling events.
API Gateway Optimization
- Implemented request caching for frequently accessed tracking endpoints
- Consolidated multiple API Gateways into shared infrastructure with path-based routing
- Optimized throttling settings to prevent cost spikes from misbehaving integrations
Infrastructure Right-Sizing
- Right-sized RDS instances based on actual usage patterns
- Implemented S3 lifecycle policies for delivery proof images
- Set up cost allocation tags across all resources for per-service visibility
- Deployed Datadog for unified observability with cost-aware alerting
Technologies Used
“Our cloud costs were eating into every delivery we made. QuantaCodes helped us understand exactly where money was going and fixed it without any downtime. The Karpenter setup alone saves us thousands per month, and our platform actually performs better during peak hours now.”
Ready to achieve similar results?
Let's discuss how we can help transform your business with the right technology solutions.