What we take on
- A foundation that will pass an audit — accounts and projects organised for separation, identity and access on least privilege, logging turned on before anyone needs it, and nothing that only one person understands.
- Migrations — off shared hosting, off a server under someone’s desk, or between clouds — planned, rehearsed and cut over with a way back.
- What you already have — the account someone set up in 2019, the DNS nobody has logged into, the backup job that has never been restored. We document it first, then fix it in the order that matters.
- The infrastructure AI needs — model endpoints, queues, and the cost and usage monitoring that keeps an experiment from becoming an invoice.
- Managed hosting — when a hyperscaler is more than the job needs, sites and applications run on Minservers, the platform we operate ourselves: per-host TLS, hardened containers, nightly integrity audits and health checks every five minutes.
Choosing where it runs
AWS, Google Cloud and Azure are all good at the same fundamentals and differ mostly in what your team already knows, what your other software assumes, and where the data is allowed to live. We’ll recommend one and say why — including, more often than you’d think, that a single well-run server is the right answer.
How we run it
Infrastructure is described in code, so it can be reviewed, repeated and rebuilt. A backup counts when it has been restored, and a monitor counts when it has alerted. Changes are planned, backed up first and verified after — against the live system, not a hunch. Every environment we operate has a runbook, and you get a copy.
What you won’t get
Kubernetes for a brochure site. A managed-services contract that hides what you’re paying for. Lock-in: it’s your account, your domain and your data, and you can walk away with all three.
