01 / week 1
Teardown
Cost & Usage Report, tagging reality check, top 20 line items ranked by recoverable dollars.
We don't just give you a report.
We ship the fix.
Forward-deployed engineers who find the waste in your AWS bill, then sit in your repo and your console until it's gone. We get paid a share of what you actually save.
aws · where the money leaks
illustrative
- EC2 on-demand
- oversized, never revisited
- RDS replicas
- running, unused
- S3 without lifecycle
- grows forever
- NAT + data transfer
- misrouted egress
- EKS nodes
- half empty
- Non-prod
- on at 3am
week 1 output your line items, ranked
The patterns we go looking for, with a dollar value put against each one in week 1. An illustration of our method, not a client bill.
Who you're working with
Engineers who have run production AWS estates in-house and carried the pager for them, not analysts who have only read the bill.
The usual engagement
- 01Six weeks of discovery calls.
- 02A 94-page findings deck.
- 03A backlog handed to your team.
- 04Nothing ships. The bill is unchanged.
Hands on keyboard
- 01Week 1: read-only access, full bill teardown.
- 02Week 2: we're in your repo opening PRs.
- 03Week 3 to 6: rightsizing, commitments, guardrails, all merged.
- 04You keep the runbooks. We invoice on what you saved.
How a six-week engagement runs
One engineer embedded in your Slack and your CI. Two if the estate is large. No account managers.
02 / week 2
Quick wins, merged
Orphaned volumes, idle replicas, forgotten environments, storage lifecycle. Where we expect to cover our own fee.
03 / week 3 to 5
Architecture work
Rightsizing with real load data, Graviton migrations, NAT and egress paths, autoscaling that actually scales down.
04 / week 6
Guardrails & handover
Budgets, anomaly alerts, cost checks in CI, and a commitment plan your team can run without us.
What we actually change
Every item below is work we do in your account or your codebase, not a recommendation on a slide.
Compute rightsizing
Instance families matched to real CPU/memory profiles. Graviton where the build allows it.
Commitment strategy
Savings Plans and RIs sized to your floor, laddered so you don't lock in a bad year.
Storage lifecycle
S3 tiering, snapshot expiry, log retention that stops paying for 2019.
Data transfer & NAT
VPC endpoints, cross-AZ chatter, CDN paths. Often the quietest six figures on the bill.
Kubernetes efficiency
Requests vs. actual usage, bin-packing, Karpenter/spot pools with real disruption budgets.
Data platform spend
Warehouse sizing and auto-suspend, partition pruning, retention on the pipeline nobody owns.
Environment hygiene
Off-hours schedules for non-prod, ephemeral preview envs that actually get torn down.
Cost in CI
Terraform diffs annotated with monthly delta, so the next regression gets caught in review.
Ownership & tagging
A tagging scheme that survives contact with reality, plus per-team showback.
Straight answer
We're new, and we've priced it that way
We're not going to show you a case study from a client we haven't had yet. What we'll do instead is find the waste in your bill for free, show you the number, and only charge you once the savings are real and measured. If we find nothing, you've lost a week of read-only access and we've lost rather more.
- $0
- to find out what's recoverable
- 5 days
- from access to findings
- 0
- changes without your review
Engagement models
The assessment is free because we'd rather show you the number than sell you a discovery phase.
Assessment
Free
5 business days, read-only access.
- Ranked savings map with dollar estimates
- Effort and risk per item
- Yours to keep, whether or not we continue
Preferred
Savings share
A share of realized savings
Agreed before we start, billed against verified reductions.
- Embedded engineer, hands on keyboard
- We open the PRs and see them merged
- Measured against a frozen baseline
- No savings, no invoice
Embedded retainer
Monthly fixed fee
For estates where cost work never ends.
- Named engineer, fixed days per month
- Anomaly response and commitment management
- Quarterly efficiency review with finance
We quote both against your actual estate, on the assessment call. The assessment itself is free either way.
The questions you should ask us
Do you need write access to our AWS account?
Not to start. The assessment runs read-only. Write access comes later, scoped, and most changes ship as PRs your team reviews.
How do you measure "realized savings"?
A baseline frozen at engagement start, normalized for growth in traffic and headcount. We agree the formula in writing before any work begins.
What if our team can't spare review time?
We sequence low-risk work first so early PRs are near-trivial to approve. Expect two to three hours a week from one of your engineers.
Is this AWS-only?
AWS is where we're deepest. GCP and Azure estates are welcome, and multi-cloud egress is often the easiest win of all.
Send us last month's bill. We'll send back the number.
Five business days, no cost, no obligation. If there's nothing worth fixing we'll tell you that too.