I build infrastructure for a living, but I think about it the way some people think about static analysis: the goal is to push correctness as far left as possible, externalize judgment into systems that don't depend on me being in the room, and let mechanical enforcement carry the load that procedural discipline can't.
Most of my career has been spent in places where the wrong abstraction was already the load-bearing one — replatforms, migrations, decommissionings, regulated environments where the cost of getting it wrong shows up in audit logs months later. I've learned to design from contracts before code, name the dangerous states before they cost anyone sleep, and write things down so the same decision doesn't have to be made twice.
I'm interested in observability platforms, data infrastructure, and AI engineering — particularly where AI is being used to sharpen human judgment rather than replace it. I find every layer of the stack has problems worth solving on its own terms.
I write occasionally about engineering decision-making and what changes when AI becomes a credible collaborator on the work itself.