I turn repetitive manual work into AI workflows that save hours every week.

I spent the last five years as Principal Engineer at Genuin, where I owned a platform that about 2.2 million people used every month across more than fifty apps. I built the React web app, the Python, Node and Go services behind it, the search and ranking, and the pipeline that shipped all of it.

The part I kept coming back to was the manual work. Content that someone reviewed by hand. Moderation queues. Reports assembled every Monday. We replaced those with LLM features and agent workflows that ran in production, and the interesting problem was never the model. It was making something reliable enough that people stopped checking it.

I also build my own things. Social Marketing AI is an AI-native system I designed and shipped on my own, and it is where I worked out how I think automation should behave: every AI provider swappable, answers grounded in real data, and a human approval step the system will not act without.

Before that I spent a decade on Apple platforms, which is still the work I am fastest at. Swift and SwiftUI, an SDK embedded in more than fifty third-party apps, Bluetooth for connected hardware, and Metal and MediaPipe for models running on the phone instead of in the cloud.

Fourteen years in. Still most interested in the gap between something that demos well and something you can leave running.