Director of Development and Product, WAND Digital

I was hired to replace one person. He had seven years in the platform, and before that he had been part of a product team of ten or more. That team was already gone. What was left was one person's memory of a decade of decisions, and a notice period.

Underneath him was a WebForms monolith over ten years old, carrying a decade of legacy code and a decade of legacy product on top of it, running the menu boards for forty national restaurant brands.

And the person at the far end of it — the manager at a brand who wants a price changed on Tuesday — could not do it. She filed a ticket, and we did it for her.

Kept it running and made it sellable. Platform maintenance, and getting sales the answers and the demos they needed to close.

That is not the interesting part of the job, but it is the part that buys the right to do the interesting part. Nobody hands a new director an architecture budget. You earn it by being the reason the quarter closed.

We were growing, so I hired two engineers — one backend, one frontend.

Then the stack, worked backwards from where the product needed to be rather than forwards from where it was.

C# API microservices on AWS. SNS and SQS between them, API Gateway in front, OAuth 2.0 for identity, React on the client. CI/CD pipelines and infrastructure as code, so a deploy stopped being an event on a calendar.

I built the infrastructure before I built the team on it. That order matters more than it sounds: engineers landing on a stack that already has a shape ship in weeks instead of arguing about it for a quarter.

Part of the reason to do it at all was a thing that did not exist yet. A monolith cannot be talked to. Services with real boundaries and an API gateway can — which is what made the MCP server possible later, and what let AI agents start handling customer support on their own.

New product management applications that generated revenue rather than absorbing it.

Self-serve menu design and publish — the first time in the platform's history that a customer could design their own menu and put it on the wall without coming through us. The ticket she used to file, she now doesn't.

An MCP server, so agents could answer customers autonomously instead of queueing behind a human.

Over my tenure the digital signage footprint doubled and so did revenue. AWS cost grew, but nowhere near in step — close to flat against the growth, while carrying three new applications and twice the footprint. The migration off the legacy stack was well underway when I left, not done. Taking a decade-old monolith apart under forty live national brands is a multi-year job, and anyone who tells you they finished one in a tenure is describing a smaller platform than this.

The hire I am most sure about was the one I argued for rather than the ones I made: a product manager. Three engineers shipping the wrong thing quickly is worse than two shipping the right thing. Adding someone whose whole job was deciding what was worth building is what made the team fast in the way the business could feel.

None of that was clean. I picked service boundaries that had to be redrawn, migrated things in an order I would not repeat, and spent money in the first year on infrastructure that was answering a question we did not have yet.

That is the part worth saying out loud, because it is where the practice comes from. I do not advise on scaling a SaaS platform from a book. I advise on it from five years of doing it to a live one with forty national brands on it, and from remembering exactly which decisions cost me a quarter.