"DevOps engineer" is one of the least precise titles in tech. It can mean the person who builds your cloud infrastructure, the person who owns your CI/CD pipelines, or the person who gets paged at 3am when production falls over. Sometimes it means all three. The most common hiring mistake is posting for "a DevOps engineer" without deciding which of those jobs you are actually filling, then being surprised when the great pipeline engineer you hired is lost in your Terraform.
Key Takeaways
- Scope the role first. Infrastructure, delivery, and reliability are different skill sets.
- Test for judgment under failure, not tool trivia. Knowing Kubernetes exists is not the same as running it.
- The best signal is how a candidate reasons about a broken deploy, not how many AWS services they can name.
- Measure the work with delivery metrics, not activity (the four keys).
The Three Jobs Hiding in One Title
Before you write the job post, pick the center of gravity. An infrastructure-leaning DevOps engineer lives in cloud provisioning, networking, and cost. A delivery-leaning one owns build pipelines, deploys, and developer experience. A reliability-leaning one owns uptime, observability, and incident response. Strong engineers cover more than one, but almost nobody is elite at all three, and a job post that demands all three reads as a company that has not thought about what it needs.
The clearest framing comes from the research on software delivery performance. DORA, the longest-running study of its kind, measures teams on how often they deploy, how fast changes reach production, how often those changes fail, and how quickly they recover (DORA). A good DevOps hire moves those numbers. Ask a candidate which of them they have personally improved and by how much.
Test Judgment, Not Tool Names
The resumes will be a wall of logos: AWS, Docker, Kubernetes, Terraform, Jenkins, Prometheus. Docker and the major clouds top the tooling in the developer survey for a reason (Stack Overflow 2025), so listing them is table stakes, not signal. What matters is what the candidate does when the tools fail.
| Weak interview | Strong interview |
|---|---|
| "Have you used Kubernetes?" | "A pod keeps restarting. Walk me through it." |
| "List the AWS services you know" | "Your cloud bill doubled overnight. What now?" |
| "Do you know Terraform?" | "Someone changed infra by hand. How do you recover?" |
| Counting years with a tool | Tracing a real incident to root cause |
A Concrete Screen
Describe a real failure and let them drive. A deploy went out an hour ago, error rates are climbing, and the dashboard shows spiking latency on one service. Do not hand them the answer. A senior asks what changed, checks the deploy history, considers a rollback before a root-cause hunt because customers are hurting now, and only then digs into logs and metrics. A weaker candidate starts theorizing about code without first asking the single most useful question in operations: what changed recently. That instinct, stop the bleeding then investigate, is the job.
The Honest Counterpoint
At an early startup, you may not need a dedicated DevOps engineer at all. A strong senior backend developer who is comfortable with infrastructure can often carry the whole thing until you have real scale, and hiring a specialist too early gives you an expensive person with not enough to do. The signal that you actually need the role: deploys have become scary, on-call is burning out your product engineers, or your cloud bill has grown faster than your usage. Before then, "a backend dev who owns infra part-time" may serve you better than a full DevOps hire.
Scoping and Cost
Once you know which of the three jobs you are filling, the interview and the rate both get clearer. A senior DevOps or platform engineer in the US often runs $150 an hour and up. Nearshore in Latin America, the same seniority lands roughly in the $60 to $95 range with real-time overlap for incident response, which matters more here than in most roles (nearshore vs offshore). Whoever you hire, hold them to delivery outcomes, and read our staff augmentation guide for CTOs if you are deciding between a hire and an augmented team. See available engineers.
Frequently Asked Questions
What does a DevOps engineer actually do?
It depends on the role's center of gravity: cloud infrastructure, CI/CD delivery pipelines, or production reliability and on-call. Decide which before you hire, because the skill sets differ.
How do I test a DevOps candidate?
Give them a realistic failure scenario and watch how they reason. Strong candidates ask what changed, consider stopping customer impact first, and trace to root cause instead of guessing.
When does a startup need a dedicated DevOps engineer?
When deploys become risky, on-call burns out your product engineers, or cloud costs outpace usage. Before that, a backend engineer comfortable with infrastructure often suffices.
How much does a senior DevOps engineer cost?
In the US, commonly $150 an hour or more. A senior nearshore DevOps engineer in Latin America runs roughly $60 to $95 an hour with timezone overlap for incident response.
The Bottom Line
Hiring a DevOps engineer starts before the interview, with an honest decision about which of the three jobs you need filled. Test how candidates reason under failure rather than how many tools they can list, hold the role to real delivery metrics, and do not hire a specialist before the pain justifies one. Scope first, then screen.
Roberto Espinoza is CEO of Ruzora, which helps US startups hire pre-vetted senior LATAM engineers, with a vetted shortlist in 72 hours. See available engineers.
