senior full-stack software engineer / applied AI
Deloitte since 2021 (FedNow, IRS.gov, Michigan UI). Before that, the ASU Mobile App. Tempe, AZ, remote-friendly.
KevinMurphy
I build software end to end: the data model, the API, the cloud it runs on, and the screen people use. A degree in anthropology keeps me watching how people really use it, and AI is part of how I build every day.
PoolRMFirm initiativeNestJS + AWS backend, AI-built, 300+ weekly users

approach
Typed contracts, bounded queries, observable failures, performance budgets that hold under traffic. Five years at Deloitte have put this discipline against real stakes. React modernizing Michigan’s unemployment insurance for 270K residents across 2,000+ pages, ReactJS + TypeScript handling 880M IRS.gov visits and $4.7T in tax collection, Salesforce LWC cutting bank onboarding to FedNow from 60 days to 7. Before any taste shows up, the build has to not break.
The user-facing layer is where a product earns or loses the user, on every interaction. A button that hesitates, a chart that flickers in, an error that says "something went wrong" instead of what to do next: those are the moments trust is decided, and they happen on the surface. Most teams treat the surface as a delivery layer; I treat it as the proof that the rest of the build works. My open-source artifacts on user-facing AI surfaces (tablesalt, streamfield) are where I show that work in public.
Senior isn’t a title. It’s the standard you set for the team. At Deloitte I coached seven juniors and mid-level consultants; one teammate went from 40% to 95% utilization inside a year. The most durable thing I’ve shipped is the bar people kept running after I rotated off the engagement.
projects
selected writing
off the clock