Senior software engineer · Healthcare
Systems that keep working when things go wrong.
I'm Nate Parker. For the past five years I've built and defended the platform at a healthcare software company — mostly Elixir and Ruby on Rails, with a security practice running underneath all of it.
Currently
Senior engineer at a healthcare startup, where I've spent five years on the same codebase — long enough to have opinions about what actually keeps software alive. Right now I'm the primary engineering resource on our push toward a top-tier healthcare security certification, which is turning out to be less about documentation than about engineering.
Outside of that: CTFs, reverse engineering, and a homelab that mostly exists so coding agents have somewhere to work while I sleep.
Writing
3 recentDeploying to the always-on box with one word
I already had a registry-based deploy: CI cross-built the image under emulation, and the Mac mini pulled latest. I replaced it with a push, a pinned commit, a native build on the mini, a health check, and a Caddy entry it writes itself.
Every dev server gets a name
localhost:4321 only works on the machine it runs on and means nothing a week later. One wildcard on the always-on box, a second Caddy on the desktop, and a thirty-line helper give every dev server a URL that works from any machine in the house.
Leaving zsh, keeping the prompt
Moving to a distro that ships its own bash setup meant either fighting it or layering on top. I layered: the prompt I missed, the aliases I actually use, autosuggestions in bash, and two bugs that only show up over SSH.