Fractional DevOps

If your team builds software and nobody owns the infrastructure, this is the arrangement most of that work gets bought under. You get an infrastructure engineer for the days you need one, rather than hiring one.

What this covers

Hands-on work, not advice. The usual shape of it:

• A deployment pipeline has stopped working, or never quite worked.
• An AWS environment needs building properly, in Terraform, so it can be rebuilt the same way twice.
• A Kubernetes cluster needs building, or the one you have is misbehaving.
• A migration needs running — observability tooling, on-prem to cloud, or between managed services.
• Monitoring is installed and nobody trusts what it says.

I work with Kubernetes, AWS, Terraform, Ansible, Helm, FluxCD, GitLab CI, and the Prometheus/Grafana/Loki/Mimir observability stack. The four areas this falls across each have their own page under services.

How time is agreed

Two ways, and they’re the same work either way.

A piece at a time. You describe what needs doing. I scope it, quote a price, and we agree dates. Most work starts here.

A block of days each month. For teams where work keeps coming up, a set number of days a month, booked as you need them. Days are agreed against work, not held on standby.

I take on a limited number of engagements at a time, so dates get agreed rather than assumed. Scope and cost are agreed before anything starts — no open-ended hourly meter, no minimum contract length.

Rate

£950–£1,100 per day, depending on the work. A day is 7.5 hours.

Fixed-price quotes for a defined piece of work are available and are usually the better option where the scope is clear. Where I quote a fixed price, that’s the price.

What this doesn’t include

• On-call cover, incident response guarantees, or any response-time commitment. That needs more than one person and I don’t offer it.
• Day-to-day operational ownership. Your team owns and runs the systems, during the work and after it. Anything I build is handed over with documentation, not kept in my head.
• Windows servers, Microsoft 365, Google Workspace, laptops, printers, phones, or general office IT. That’s a managed IT provider’s work, not mine.

If continuous cover is what you’re after, a managed service provider is the right thing to buy and I’d rather say so now than three months in.

If you’re not ready to hand over work

There’s a smaller way to start: a fixed-scope observability audit — a review of what telemetry you collect, whether the alerts mean anything, and what the stack costs to run. Delivered as a written report, a prioritised list of what to fix with effort estimates, worked configuration examples you can actually use, and a call to go through it.

£2,500–£3,500, typically five working days, price agreed before it starts.

It’s a way in rather than the main thing on offer. Most people are better served by just having the work done.

How to start

Get in touch with roughly what you need — the contact form or email both work. A first conversation costs nothing and commits you to nothing. If I’m not the right person for it, I’ll say so.

Remote by default. London on-site where it’s genuinely needed.