Offers
Platform & production readiness
From “works on our machine” to reliable operations: calm deployments, visible failures, clear ownership.
“Every release is an event — and we notice failures when customers call.”
- Format
- 4–10 weeks depending on the starting point
- Who it is for
- Teams that want to ship faster but get stuck in operations
How you notice it
- Releases never happen on Fridays — and in the evening only with standby.
- You learn about outages from support, not from monitoring.
- A rollback is theoretically possible, but nobody has ever practised it.
- The environment grew by hand over years and exists exactly once.
- There are plenty of alerts, but most get dismissed.
Outcome
- Kubernetes setup or clean-up, infrastructure as code
- CI/CD with reproducible builds and safe rollbacks
- Observability: logs, metrics, traces and meaningful alerts
- Production readiness check with a prioritised backlog
How we work
01
Week 1 — readiness check
We review build, deployment, configuration, data, monitoring and emergency paths along a fixed list. The result is a backlog sorted by risk, not a wish list.
02
Weeks 2–4 — shipping without dread
Reproducible builds, automated deployment, safe rollbacks and environments as code. We rehearse the fallback instead of only documenting it.
03
Create visibility
Logs, metrics and traces reduced to what matters, plus a few alerts that are allowed to wake someone. Dashboards follow the questions asked during an incident.
04
Handover & operating routines
Runbooks, an on-call model and one guided incident. Afterwards the team runs operations without us.
Frequently asked
- Do we need Kubernetes for this?
- Not necessarily. For many teams containers on a managed platform are the calmer choice. We clean up existing clusters — or advise against them when the effort outweighs the benefit.
- We already have a pipeline. Isn't that enough?
- Having a pipeline is different from trusting it. We check reproducibility, rollback, secrets and test coverage, and start where it is unsafe.
- What does “production ready” mean concretely?
- A deployment is possible at any time without a special slot, failures are visible within minutes, there is a rehearsed fallback path and a named owner.
- Could we do this internally?
- Often yes — usually the time is missing, not the skill. That is why we work inside the team rather than next to it, and hand over step by step.
- What does it cost?
- The readiness check is small and gives a solid basis after one week. Scope is then cut together into four to ten weeks of work.
When this does not fit
If the product has no real users yet, platform work is premature. A simple deployment is enough, and the time belongs to the domain.
Often combined with
Background in the article Production readiness checklist: 12 questions before the next deployment.