Finding the 40% of the cloud bill nobody was using
Performance and cost are usually the same investigation, run twice.
The problem
Critical API endpoints were slow, and the cloud infrastructure bill was bigger than the workload actually needed.
What I did
- Profiled and optimised the slow queries behind the worst-performing endpoints.
- Worked with the DevOps team to rework resource allocation across the cloud estate.
- Drove a JavaScript to TypeScript migration, so a whole category of defect stopped reaching production.
Average response time on critical API endpoints dropped by 50%, cloud infrastructure cost fell 40%, and the TypeScript migration lifted developer productivity by an estimated 30%.
The outcome
