How to Hire a DevOps Engineer: A Step-by-Step Guide for Small Hiring Teams
Companies hire a DevOps engineer when deploys have become scary and the cloud bill has become surprising. This guide is the process for hiring someone who will make production boring, and for avoiding the common outcome of hiring a tools enthusiast who makes it more complicated.
- Name the pain, not the platform: 'Deploys take an hour and fail one time in five; nobody knows what the AWS bill is for' is a role. The role description is the first screening step, and it decides who applies.
- Decide whether this is a platform role or an operations role. Building internal tooling so developers can ship safely is a different job from running what exists and being on call for it.
- Source where devops engineers already are: backend engineers at your stage who have been the de facto operations person, and site reliability and platform engineers at larger companies who want to own the whole stack.
- Screen every CV against a written rubric of the four things the job cannot be done without, then interview for evidence: what the candidate built, decided or fixed, scored against notes written before the interview.
- Decide within a week of the final interview and make the offer with reasons. The common mistakes below, starting with hiring for a tool list and getting a migration nobody asked for, are how good candidates are lost.
The process
- 1
Name the pain, not the platform
'Deploys take an hour and fail one time in five; nobody knows what the AWS bill is for' is a role. 'Kubernetes expert' is a wish. The first attracts engineers who fix things; the second attracts engineers who install things.
- 2
Decide whether this is a platform role or an operations role
Building internal tooling so developers can ship safely is a different job from running what exists and being on call for it. Small companies need both from one person for a while; say so, and hire for the operations half first.
- 3
Source from engineers who have been on call
The best candidates are backend engineers who drifted into operations because nobody else would, and operations engineers who learned to code because it was faster. Both write about incidents on their CVs.
- 4
Screen against a rubric, not a feeling
Write down the four things the job cannot be done without and score every CV against them before reading any further. Three minutes per CV with a rubric beats fifteen without one, and the decisions can be explained afterwards.
- 5
Interview for evidence
Every interview question should ask for something that happened: a system, a deal, a decision, a failure. Score answers against what a strong one contains, written down before the interview, so two interviewers agree.
- 6
Decide fast and tell everyone
Candidates judge a company by its process. Decide within a week of the final interview, tell the ones you passed on, and make the offer with the reasons you chose them. The best candidates have other offers.
Where to find candidates
- Backend engineers at your stage who have been the de facto operations person
- Site reliability and platform engineers at larger companies who want to own the whole stack
- Referrals from your engineers, who know who fixed the last outage at their previous company
- Communities around your cloud provider and your orchestration tooling
- A posting that describes the current deploy process honestly, failures included
Mistakes that lose good candidates
- Hiring for a tool list and getting a migration nobody asked for
- No on-call rotation or incident process, so the new hire is on call alone forever
- Letting the DevOps engineer own everything operational, so developers stop learning it
- Measuring on uptime alone and getting a hire who blocks every change
- Ignoring security basics in the interview and finding out at the breach
Running it well
Interview with the pipeline and the incident. Ask what they changed about how software reached production at their last job, with numbers, and ask for the timeline of their worst incident. Candidates who improved a pipeline can quantify it; candidates who ran an incident well can narrate it. Everything else on the CV is inventory.
Ask about security in the interview, seriously: secrets, access, rotation. Small companies are breached through the basics, and this hire is the person who prevents that. A candidate who cannot say where secrets live and how they rotate is not ready to own production.
Write the on-call expectation into the offer, with the rotation and the compensation for it. The fastest way to lose a good operations engineer is to make them the only one who ever gets paged.
Frequently asked questions
Should our first hire be DevOps or a backend engineer who can do it?
If deploys and infrastructure are the bottleneck, a DevOps engineer. If features are the bottleneck and infrastructure is merely untidy, a backend engineer with operations experience, and revisit in a year.
Do we need Kubernetes experience?
Only if you run Kubernetes and intend to keep running it. Many small companies would be better served by simpler platforms, and a good candidate will say so if asked.
How do we assess security knowledge without a security specialist?
Ask the basics: where secrets live, how access is granted and removed, when credentials were last rotated, what would happen if a laptop were stolen. Good answers are concrete; bad ones are reassuring.
Related guides
Screening devops engineer CVs this week?
Paste the job description into WadeCV, check the rubric it drafts, and upload the CVs. Each one is read against the rubric and scored with an explanation. Ten screenings are free; there is no subscription.
Screen your first CVs free