Case Study
Serverless cost & reliability uplift
A SaaS product had unpredictable latency and rising AWS costs. I re-architected the serverless stack to improve resilience and spend efficiency without slowing delivery.
Highlights
- Costs reduced by 32% through async offloading and right-sized concurrency.
- p95 latency stabilized under 400ms with provisioned concurrency on critical paths.
- Safer deploys via canaries and automatic rollback tied to CloudWatch alarms.
What I built
- Event-driven workflow using SQS + Lambda for bursty workloads and idempotent handlers.
- Infrastructure-as-code with CDK, including API Gateway configs, WAF, and observability defaults.
- Cost/perf dashboards in CloudWatch and Grafana to track business KPIs alongside infra metrics.
Impact
- Developers shipped weekly without fear of regressions thanks to staged rollouts.
- Finance gained predictability with tagged spend and automated budget alerts.
- Customer-facing reliability improved, reducing support tickets tied to timeouts.