Keep the infrastructure standing
Deployments without surprises
Your cloud, containers and deployments put in order and defined as code, with backups that genuinely restore and alerts that arrive in time.
The problem it solves
Infrastructure works until it doesn’t. There is usually a machine nobody dares to touch, backups that have never been restored and a deployment only one person knows how to run.
What I do
I review what is there and leave it in a state that holds: architecture on one cloud or several, Kubernetes and containers, infrastructure defined as code with Terraform and Ansible, reproducible deployments, backups tested by actually restoring them, monitoring that alerts you before a customer notices, and basic hardening of everything exposed to the internet.
I also prepare the platform for AI workloads: pipelines that take a model to production, version it and watch it, so that each deployment is not a new project.
Whatever is set up is documented for your team, with the access in your name.
Scope
- Included: an audit of the current environment, a plan prioritised by risk and delivery of what is agreed.
- Not included: 24/7 on-call cover. If you want hours every month afterwards, there is maintenance.
Deliverables
- A written initial audit, with risks ranked by impact.
- The infrastructure defined as code in your repository.
- A reproducible deployment pipeline that more than one person can run.
- Backups with a real restore tested and documented.
- Monitoring and alerts on what genuinely affects the service.
- Runbooks for routine tasks and common incidents.
How it is measured
Recovery time proven with a real restore, the number of manual steps left in a deployment, how many people on the team can run it, and how much of the infrastructure is defined as code. Figures you can look at on day one and again after three months.
What I need from you
Administrative access to the environment and knowing what can be stopped, and when.