What we do
Tools
Industries
Department
How we workLet's talk →
Home/DevOps Engineer
DevOps

What a DevOps Engineer Does Between Git Push and Production

Every software team has a path from a developer saving code to customers using it. A DevOps engineer makes that path repeatable, so a release works the same on a Friday afternoon as it does with its builder watching. Hiring one? First work out whether you need a person, a project, or one fix.

The path a commit takes, and where it tends to break

Studying for a DevOps role? Each stop here shows up on job listings. Running a team? Mark every step that still needs a particular person to do it by hand.

  1. 1

    Code is pushed and reviewed

    git push pull request merge

    Version control, branch rules and code review decide what's allowed to reach the main branch. This is the cheapest place to catch a bad change.

    Watch for: merges straight to main with no review and no tests attached.

  2. 2

    A pipeline builds and tests it

    Continuous integration (CI) tools such as GitHub Actions, GitLab CI or Jenkins build the app and run its tests on every change, so nobody has to remember to.

    Watch for: tests switched off "just for this release" that never get switched back on.

  3. 3

    Infrastructure is written down as code

    Servers, databases and networks are described in files (Terraform, for example) instead of set up by clicking around a cloud console. That's what lets someone rebuild them, or review a change before it happens, on AWS or Azure.

    Watch for: settings changed by hand in the console that never made it back into the code.

  4. 4

    Secrets and settings are handed to the app safely

    Passwords, API keys and per-environment settings come from a secrets store with access control, not a file copied between laptops.

    Watch for: a production password that only exists in one person's local setup.

  5. 5

    It deploys, with a way back

    Continuous delivery (CD) pushes the tested build out, ideally a little at a time, with a rollback that has actually been tried.

    Watch for: a rollback plan that's only ever been described, never run.

  6. 6

    Someone is told when it breaks

    Logs, monitoring and alerts that reach a person who can act. This is the step most small teams skip, and the one customers notice first.

    Watch for: alerts going to a shared inbox nobody reads, or to someone who left.

A full-time DevOps engineer, a project, or just a fix?

The job title is the same in every case. What changes is how much ongoing attention your setup needs.

Where your team is now What usually fits Ask yourself
A few developers, one app, releases done by hand a couple of times a month A one-off project: script the deploy and put the infrastructure in code, then your developers run it Who on the team will own the pipeline once it exists?
Several services, releases every week or every day, outages that cost real money A full-time DevOps engineer, or a team that runs it with you ongoing Can one hire carry on-call alone without burning out, and who covers their holidays?
Built by a contractor who's gone, and nobody is sure what runs where A takeover: document what exists first, then fix it Does your company own every cloud account, repository and domain login?
You need someone working inside your dev team for months Staff augmentation, a person rather than a project Who sets their priorities day to day?
One step of the pipeline is broken and the rest is fine A short fix. One broken CI step is often a half-day job, not a project. Is it really one step, or the first symptom of a setup nobody understands?

What we usually find when we take over someone else's pipeline

None of these stop the app running today. They're why the next change, or the next outage, takes days instead of minutes.

The console and the code disagree

How you'll know: Running the infrastructure tool wants to change or delete things nobody touched on purpose.

Fix: Bring the code back in line with what's actually live, then stop hand edits in the console so it stays that way.

Deploy keys tied to a contractor's GitHub login

How you'll know: The pipeline authenticates with a token from someone's own GitHub account, or a production settings file was once pasted into a chat.

Fix: Rotate every key and token first, since you can't know who still holds the old ones. Then issue new ones from accounts and a secrets store the company controls.

Backups exist, but nobody has restored one

How you'll know: No one can say how long a restore takes, or whether last night's backup is complete.

Fix: Restore into a scratch environment and time it. Do it again every few months.

A cloud bill no one can explain line by line

How you'll know: Servers named things like "test-old-2" that have been running for a year.

Fix: Tag every resource with an owner and a purpose, check what's truly unused, then shut it down.

Alerts wired to people who are no longer around

How you'll know: You hear about an outage from a customer, not a page or a message.

Fix: Route alerts to a rota your team owns, and send a test alert to prove it lands.

Questions people ask

What does a DevOps engineer do day to day?

Keeps the route from code to production working: maintaining CI/CD pipelines, changing infrastructure through code, handling secrets and access, watching monitoring and alerts, and joining the call when a deploy goes wrong. In small companies the role often includes general cloud administration too.

Do we need a full-time DevOps engineer or a project-based fix?

It depends on how much ongoing attention your infrastructure needs. A team releasing a few times a month usually needs a project and an owner afterwards. A team releasing daily across several services usually needs someone on it full time. n-frames takes on either: a scripted, documented project your team then runs, or staff augmentation when a person is what you really need.

Can you take over infrastructure someone else set up and then left?

Yes, it's one of the most common jobs n-frames gets. We write down what's actually running, and who has access to it, before we change anything. The undocumented part is usually the real problem, not the servers themselves.

Is this the same as hiring a cloud engineer?

They overlap. DevOps centers on the pipeline and the release process. Cloud engineering centers on the infrastructure itself: networks, accounts, costs, security. In a small team one person often covers both. Tell us what's broken and we'll scope the right kind of work.

What should someone learning DevOps focus on first?

Linux and the command line, Git, one CI tool, one cloud provider, and one infrastructure-as-code tool. Then build a small app that deploys itself with tests and a working rollback. Being able to explain why each step exists counts for more in interviews than a long list of tools.

Let's talk

Tell us the one thing your team does manually that eats up time. We read every message and reply within a day.