HireNimbus

I've built most of the core engineering at HireNimbus from the ground up: two React Native apps—the pros app for service professionals and the homeowner app for the marketplace—a full Supabase-to-AWS infrastructure migration, and an AI-powered quotation engine that went through three model providers before the right one stuck. Alongside that, I co-designed a meta-engineering harness with our CEO, a system for making AI-generated code reliable and auditable rather than something you take on faith. It is now a published paper, which says more about how common the underlying problem is than about how clever the solution was.

On payments, I built the Stripe layer end to end. Service pros onboard through Stripe Connect; homeowners pay deposits and final invoices via checkout sessions and payment intents, with funds routed to the pro's connected account. Pros subscribe to the platform through Stripe Checkout, and webhooks keep subscription state in sync—trial periods, renewals, and status changes. Getting deposit arithmetic right on final invoices, reconciling partial payments, and making sure webhook failures do not leave jobs in ambiguous states took more iteration than the happy path suggested.

Most of what I've learned here comes down to one thing: the interesting work usually is not writing the code. It is figuring out what you actually need to trust before you ship.