Linux, self-hosting & NixOS
The layer under the application: reproducible NixOS hosts, self-hosted services owned outright, and build environments that behave the same on every machine.
A web product is only as dependable as the machine it runs on. Nixperts is comfortable all the way down, and would rather own that layer than hope somebody else is holding it.
What this looks like
- NixOS hosts described entirely in code: the same configuration produces the same machine, and a bad deploy rolls back to the previous generation.
- Reproducible development environments with Nix flakes, so “works on my machine” stops being a category of bug.
- Self-hosting: putting the services a business depends on — Git, CI, monitoring, databases, internal tools — on infrastructure it controls.
- The boring essentials done properly: TLS, reverse proxies, backups that have actually been restored from, and logs somebody will read.
- Migrations away from managed platforms when the bill or the lock-in has stopped making sense.
On Nix, honestly
Nix has a real learning curve, and it is not the right answer for every team. It pays off when a machine has to be rebuilt, audited or handed over and nobody wants to discover what was configured by hand two years ago. Expect a straight answer about which of those situations applies.