Senior DevOps for startups too small to hire one.
We look after AWS for seed and Series A teams with 2 to 15 engineers. Prices are on the page, the first piece of work is a five-day review, and you can stop any time after that.
Not ready to talk? The runway calculator shows what your AWS bill does to cash once credits run out.
Most founders call us at one of five moments.
- Readiness review
You're about to launch, or just did
The app works. Nobody is sure what happens when the database fills up, or who can still log in as root.
- Cost review
Your AWS credits end in a few months
The bill says $0 today. The real number behind it is the one your runway has to survive.
- SOC 2 infrastructure
An enterprise customer asked for SOC 2
The security questionnaire is sitting in someone's inbox and the auditor will want to see your AWS account.
- Heroku to AWS
You're still on Heroku
It works fine. Heroku stopped building new features in February 2026, and you'd rather move on your own schedule.
- Fractional DevOps
Infrastructure is eating a developer's week
Someone senior keeps getting pulled off the product to fix deploys. You need the work done, not another hire yet.
Who we're good for, and who we aren't.
Good fit
- 2 to 15 engineers, no DevOps person yet
- Running on AWS, or about to be
- Seed or Series A, with real customers on the system
- You want to own the code and the accounts at the end
- US or UK hours, or happy to overlap with them
Not a fit
- You need a staffed 24/7 operations centre
- Mostly on-premise or bare metal
- You already have a platform team of five
- You want a body in a chair for 40 hours a week
- Pre-product, with nothing running yet. Call us in six months.
Fig. 1
A typical start, drawn as a runway.
What happens after you say yes.
Day 1. You create a read-only IAM role from a short Terraform snippet we send, and add us to the repo. We never ask for passwords or root.
Day 3. A 20-minute call to check we've understood the system, and to flag anything that can't wait for the report.
Day 5. The report: what's fine, what's risky, what it costs you each month, in the order we'd fix it. You keep it whether or not you hire us.
Day 30. If you move onto a monthly plan, the first fixes from the report are merged as pull requests your team reviews. The review fee comes off your first month.
Written down, so you can compare us before you talk to us.
Every engagement, as a strip.
Buff strips are fixed price, blue are monthly, pink is compliance. Prices in USD, excluding tax. Full detail on the pricing page.
Fig. 2
Page 3 of a review. The startup is made up; the findings are ones we see all the time.
What you get on day five looks like this.
ACME LEDGER LTD · AWS readiness review · p.3 of 14
Findings, in the order we'd fix them
Automated snapshots are on, retention is 1 day, and nobody has restored one. Fix: 7-day retention plus a monthly restore drill. About 2 hours.
Last used 140 and 211 days ago. Fix: remove them today, then move people to IAM Identity Center. 1 hour now, half a day for the move.
About $99 a month before any traffic, for an environment used during office hours. Fix: one gateway, or none with VPC endpoints. Saves roughly $66 a month.
The production deploy script runs from the CTO's machine with long-lived keys. Fix: a GitHub Actions workflow with OIDC. About a day.
We started OpsRunway because the same conversation kept happening. A founder with eight engineers, a product people pay for, and an AWS account that grew by accident. Nobody on the team wants to be the DevOps person, and a full-time hire costs more than the problem does, for now.
So we sell the middle option and write the prices down. A short review first, so you can see how we think before you spend real money. Then as much help as the month needs, and a clean handover when you're big enough to hire your own platform engineer. That day is a good outcome, not a lost client.
The OpsRunway team
P.S. If you only want a second opinion on one thing, an IAM policy or a Terraform module, ask. Quick answers are free and come with no follow-up sequence.
The awkward questions, answered.
What happens if you get hit by a bus?
Everything we build lives in your AWS accounts and your Git repos, as Terraform and plain documentation. There is no tooling of ours to switch off. Any competent engineer can pick it up, and our runbooks are written for someone reading them at 3am.
Do you want root on our AWS account?
No. The review needs a read-only role, which you create and can delete the minute we're done. On a plan we work through roles with only the permissions the job needs, and every change arrives as a pull request your team can read first.
Is this actually cheaper than hiring someone?
Until the work is full-time, usually yes. A US DevOps engineer is around $130k to $155k in base salary before taxes, benefits and the recruiter. Our middle plan is $4,900 a month. Our hiring guide has the full comparison, including when hiring is the better call.
Can we stop?
Plans run for three months and then month to month, with 30 days' notice. When you leave, you get the credentials, the documentation and a written list of anything half-finished.
We're on Google Cloud or Azure. Can you still help?
Most of our work is AWS, and our prices assume AWS. We do take GCP and Azure work, but we'll quote it after a call rather than pretend the strip above applies.
Can we pay nothing up front for the AWS cost work?
Yes. On the share-of-savings option we make the changes, measure the saving against a baseline agreed in writing, and invoice 20% of what you actually saved for 6 months. If nothing is saved, there's no invoice.