DevOps is four jobs sharing a title, and most disappointing hires are a mismatch rather than a bad engineer. Build and release automation, cloud infrastructure, production reliability, and platform tooling for other developers are genuinely different specialities with different strengths. Deciding which you need is most of the work.
The four jobs, and which one you actually have
If your problem is that releases are slow and manual, you need build and release automation. If it is that infrastructure is undocumented and changes are made by hand, you need infrastructure as code. If it is that the system falls over and nobody knows why, you need reliability and observability. If it is that developers are blocked waiting on one person, you need platform tooling. Write down which sentence describes you before writing the job advert.
Test incident behaviour, not tool familiarity
Tool lists are the least informative part of a DevOps interview because tools are learnable and change constantly. Ask instead about a real outage: what happened, how they found out, what they did first, what the actual cause turned out to be, and what changed afterwards. Listen for whether they distinguish the trigger from the cause and whether the fix was systemic or a patch. That story predicts their value better than any certification.
Security is inseparable from this role
Whoever builds your pipelines and infrastructure decides how secrets are handled, how access is granted and what is logged. The NIST Cybersecurity Framework gives a shared vocabulary for identify, protect, detect, respond and recover, and is worth using as the structure of the conversation. Practically, ask how they manage secrets, how access is granted and revoked, what is logged and retained, and how they would detect credential misuse.
Guard against building a single point of failure
The failure mode of this hire is a person who builds an elegant system only they understand, leaving you less able to operate than before. Make documentation and knowledge transfer explicit expectations from the start, require infrastructure to be described in code in your repository, and ask candidates how they hand over. Someone who talks naturally about making themselves unnecessary is worth more than someone who talks about what they can build.
Questions people ask about hire devops engineers
What should a DevOps job description say?
Which of the four jobs you need: release automation, infrastructure as code, reliability and observability, or platform tooling for developers. A description listing every tool selects for people who list tools rather than for the speciality you need.
How do we interview for DevOps?
Ask about a real outage end to end: how they found out, what they did first, the actual cause, and what changed afterwards. Listen for the distinction between trigger and cause, and for whether the fix was systemic.
How do we avoid depending on one person?
Require infrastructure described in code in your repository, make documentation and handover explicit deliverables, and prefer candidates who talk about making themselves unnecessary over those who talk about what they can build.