Find the money your AWS bill is wasting

A fixed-price audit for startups spending real money on AWS without a FinOps hire. You get a prioritized savings report where every finding has a dollar number and the effort to capture it, including the AI and inference spend nobody on your team has time to watch.

📅 Book a free 20-minute teardown

Who this is for

You spend $20k to $500k a month on AWS.
Nobody owns cost. It's a founding engineer's 5% job, and it shows.
Your AI or inference spend is real and growing.

If your spend is mostly one big commitment-eligible workload, you don't need me; automated savings-plan tools handle that for cheap, and I'll tell you so on the teardown call. The audit earns its fee where the waste is architectural, or in AI workloads no tool understands yet.

What you get

A savings report ranked by dollars saved per unit of effort, split into three buckets, plus one specific number: here's what I found per month, and here's how much of it you can capture this week.

Free money

Idle and orphaned resources, unattached volumes and IPs, gp2 to gp3 migrations, old-generation instances, commitment coverage gaps, S3 lifecycle policies. Fixes you can ship this week.

Architectural

Rightsizing, NAT and egress charges, data transfer paths, autoscaling gaps, over-provisioned databases. Bigger wins that need engineering judgment, which is why automated tools skip them.

AI and inference

Batch vs. on-demand inference, prompt caching, model routing, endpoint choices, GPU rightsizing. Usually the fastest-growing line on the bill and the least watched.

The report ends with a choice: implement it yourself with your own team, or have me do it and keep watching for the next leak. See a sample report generated by my open-source engine.

The part nobody else is watching: your inference bill

General cost tools treat AI spend as just another line item. AWS documents the levers; almost nobody pulls them:

Batch inference on Bedrock is priced at half of on-demand.
Prompt caching cuts up to 90% of input cost on the cached prefix.
Intelligent prompt routing saves up to 30% by sending easy calls to cheaper models.
Serverless and async endpoints scale to zero instead of idling between requests.

Those "up to" figures are AWS's ceilings, not promises. The audit quantifies what each lever is worth on your actual usage before you change anything.

Why me

I've run my own production AWS stack since 2023: a multi-tenant SaaS ingesting 3M+ rows a day across Lambda, SQS, and Aurora Postgres. I re-architected its ingestion to run 16x faster and cut the infra bill in half while doing it. I also cut its CI spend by about 60%. I've been the person on call for the bill I was cutting.

My audit tooling is open source: cost-engine, a FinOps engine that reads an AWS Cost & Usage Report and flags dollar-quantified savings, with the assumption behind each number stated.

How it works

1
Grant access
A read-only, billing-scoped IAM role plus access to your Cost & Usage Report. I never need write access to anything.
2
Kickoff call
60 to 90 minutes so I understand your architecture and what you're scared to touch.
3
Report and walkthrough
Within two weeks you get the prioritized report and a 30-minute walkthrough call. Not a PDF over the wall.

$5,000, fixed

No hourly meter, no scope creep. If the audit finds one mid-size rightsizing win, it has paid for itself.