Hands-on Staff engineering
I get difficult software work unstuck.
I work across product, architecture, code, APIs, infrastructure, security, and operations to find what is actually preventing a system or feature from moving forward. I use AI aggressively to accelerate the work, but engineering judgment still owns the result.
The useful part is knowing the difference between a problem AI can compress and a constraint the organization still has to resolve.
Recently
Four repositories. One feature. The code wasn’t the whole problem.
A recent production feature crossed four repositories. I used an agent to inspect the implementation surface, related requirements, previous behavior, API and data paths, and dependencies, then turn that evidence into a bounded implementation plan and explicit open questions.
That accelerated the engineering dramatically. It also exposed the real constraint sooner: some remaining behavior had never actually been decided, and parts of delivery depended on sequencing and ownership outside the implementation itself.
More prompting would not solve that.
This is how I use AI in practice: compress the searchable and implementable work, expose assumptions early, and get to the point where the remaining human decision becomes visible.
AI gets me to the constraint faster. It does not make the constraint disappear.
Receipts
This isn’t just a theory of work.
- 15 repositories
Critical security findings eliminated across the affected repositories and related container images.
- 22 → 14 hours
Bulk-processing runtime reduced through profiling, distributed messaging, compression, caching, and code-path work.
- Working software
ArcadeGhosts: a deployed TypeScript/Next.js application I designed, built, and operate end to end.
Code you can inspect
ArcadeGhosts is not a demo repository built for a job search. It is a deployed application I own end to end: TypeScript, Next.js, server/API code, Postgres, migrations, authentication and admin workflows, storage, caching, Playwright tests, CI, deployment, and AI-assisted engineering.
If you’re reviewing me at Staff level, the interesting part is not one clever component. It’s how the pieces fit together as one owned system.
Looking for
I’m looking for substantial hands-on engineering with real technical ownership. The title matters less than whether I can help understand, shape, build, and improve the system.
If your team has a difficult problem that crosses boundaries, I’d probably rather hear about that than read another job description.