I worked as a DevOps engineer at YesSoft before founding Cyberv, and the single biggest myth I want to retire is the idea that DevOps is a stack. It is not. It is a way of thinking about software where the people who build it are the same people who run it.

The toolkit is the easy part

Docker. Kubernetes. Terraform. Argo. Prometheus. You can install all of them in a weekend and still have a worse engineering culture than a team running a single Rails app on a Hetzner box. The tools are useful only when they encode a culture you already have.

What actually moves the needle

1. Developers carry the pager. The fastest way to make code reliable is to make the people who wrote it the people who get woken up. It changes how they design.
2. Every change has a rollback. If you can't revert it in under five minutes, you didn't ship a feature, you shipped a hostage.
3. Observability before features. If you can't see what your code does in production, you're not running software, you're running rumours.
4. Boring infrastructure. Choose technology your team can debug at 3 AM with a fever. Excitement is not an SLO.

The CI/CD pipeline is the team's contract

Read your CI pipeline. It tells you exactly what your team values. If "it must lint" is enforced and "it must have a test" is not, your team values style over correctness — and your dashboards will eventually agree.

What I install on day one

• One CI provider, one deployment target, one observability stack — the simplest possible.
• An on-call rotation, even if only two people are on it.
• A status page, even if it's static.
• A weekly incident review, even with zero incidents — to build the habit.

Everything else can wait until it justifies itself.