How it works
No discovery phase, no steering committee. Week one we read the bill. Week two we're opening pull requests.
We pull twelve months of Cost & Usage Report data, map it to services and teams, and rank every line item by recoverable dollars against effort and blast radius.
Orphaned volumes and snapshots, idle replicas, dev environments running at 3am, log retention set to forever. Low risk, fast to review, and the first place we expect to cover our own fee.
Rightsizing against real load data, Graviton where the build allows, VPC endpoints and egress paths, Kubernetes requests versus actual usage, then a Savings Plan ladder sized to your floor.
Budgets and anomaly alerts wired to your on-call, cost deltas annotated on Terraform plans in CI, and a tagging scheme your team can maintain. The goal is that you don't need us again.
Cost work touches everything. The way to do it safely is to be boring about process.
Console clicks don't survive the next deploy. Changes land in your IaC, reviewed by your engineers, on your branch protection rules.
Non-prod before prod, one service at a time, with a rollback in the same PR. Anything that touches a customer path gets a staged rollout and a watch window.
Read-only for the assessment. Write access only where work is happening, revoked at the end of the engagement.
If a saving costs latency or headroom, it goes on the list with the tradeoff written down. You decide, not us.
2-3 hrs
of your engineers' time per week, mostly code review
Week 2
when we aim to have the first saving merged
0
slide decks produced