A fractional DevOps team for startups that aren't ready to hire one.
A fractional DevOps engineer is a senior person you get for a set number of hours a month, instead of a salary. You get the infrastructure work done without a recruiter, a notice period or someone sitting idle between incidents.
Hours reset each month. Unused hours don't roll over, which keeps the price flat for both of us.
Three plans, priced by hours and how fast we answer.
| Essentials | Growth | Scale | |
|---|---|---|---|
| Price per month | $2,900 | $4,900 | $7,900 |
| Engineering hours | up to 20 | up to 40 | up to 80 |
| Production incident response | Next business day | Within 4 business hours | Within 1 hour, in support hours |
| Monthly report (changes, incidents, spend) | Yes | Yes | Yes |
| AWS cost check, savings applied | On request | Every month | Every month |
| Backup restore test | On request | Quarterly | Quarterly |
| Kubernetes operations | No | Small clusters | Yes, including upgrades |
| SOC 2 controls kept in shape | No | No | Yes |
Not sure which? Most teams with one production environment and weekly deploys start on Growth. If your first month is mostly fixing what the review found, start with Essentials plus a fixed-price project instead of paying for hours you won't need later.
What a fractional DevOps engineer actually does in the first month.
Less than you'd think is new tooling. Most of the first month is making the existing setup boring:
- Every person logs in through IAM Identity Center with MFA. Old IAM users and long-lived keys go.
- Infrastructure that was clicked together in the console is imported into Terraform, so it can be reviewed and recreated.
- Deploys run from CI with short-lived credentials, not from a laptop.
- Alerts go to a channel someone reads, for the five things that actually take you down.
- Backups exist, are retained long enough, and one has been restored on purpose.
- A budget alert fires before the AWS bill surprises anyone.
After that, the hours go where the product needs them: a staging environment that matches production, a database upgrade, preview environments, or the first steps towards SOC 2. If someone has proposed Kubernetes, our ECS vs EKS guide is the conversation we'd have first.
Salary figures: ITJobsWatch (UK, 6 months to Oct 2026) and US job-posting medians. See the hiring guide for sources.
Fractional, full-time, freelancer or agency.
| Fractional (us) | Full-time hire | Freelancer | Big agency | |
|---|---|---|---|---|
| Typical cost | $2,900 to $7,900 a month | $130k to $155k base (US), £70k median (UK), plus overheads | £525 a day (UK median), $100 to $165 an hour (US) | Quoted per project, often with a minimum |
| Time to start | About a week | 2 to 4 months to hire | Days, if they're free | Weeks |
| Cover when they're away | Yes, a small team | No | No | Yes |
| Context on your system | Builds over months, written down | Deepest, once settled | Leaves with them | Varies by who's assigned |
| Best when | The work is real but not full-time | Infrastructure is a full-time job | One defined task | A large, fixed project |
When to stop paying us: once infrastructure work fills a full working week, every week, for a few months running, hire. We'll help write the job description and interview, and hand over properly.
What we'll do, and what we won't.
On a plan
- Terraform for AWS, in your repo
- CI/CD on GitHub Actions or GitLab CI
- ECS, Fargate, Lambda, RDS, and EKS where it earns its keep
- Monitoring, alerting and on-call runbooks
- Cost checks and the fixes that follow
- Security basics: access, secrets, logging, backups
Not on a plan
- Writing your application code
- 24/7 on-call without an agreed add-on
- Desktop IT, laptops, Google Workspace admin
- Being the only people who understand your system
- Signing your SOC 2 report (an auditor does that)
Before you sign up for a plan.
Do we have to do the readiness review first?
No, but most teams do. It gives both sides a list of what the first months are for, and its $1,900 fee comes off your first month if you start a plan within 30 days.
What if a month needs more than the plan's hours?
We tell you before it happens, then either quote the extra work in writing or you move up a plan for that month. You never get a surprise invoice.
Who will we actually be talking to?
A named engineer who owns your account, with a second engineer who knows it well enough to cover holidays and sick days. You talk to them in a shared Slack channel, not through an account manager.
Can you be on call at night?
Out-of-hours on-call is an add-on, quoted once we've seen how noisy your alerts are. We usually reduce the noise first, because paying for a pager that goes off for nothing is a waste for both of us.
What happens when we hire our own DevOps engineer?
We overlap for a month: shared on-call, a walkthrough of every repo and runbook, and a list of what we'd do next if we were staying. Then you stop the plan with 30 days' notice.