Scaling the $100M+ lifecycle engine for self-serve
Moved over to run B2C marketing ops: scaling the engine via global and program-level holdouts, optimizing the AI decisions layer for self-serve, and building the repeatable acquisition playbook for M&A.
$100M+
Lifecycle revenue surface
Ops people don't ship infra. I shipped Meridian.
I moved to B2C in January. Different beast — self-serve audience, much higher volume, an AI decisioning layer picking what each user sees next. The question stopped being 'can we measure this' and became 'can we govern it at scale, and prove incrementality while we do.'
The bottleneck wasn't strategy. It was visibility. We couldn't see inside our largest-impact programs in real time, and errors compounded before anyone noticed. So I built Meridian — a governance and monitoring system for those programs — in Claude Code, Cursor, GitLab, and Vercel. Self-taught the whole stack.
Errors are down. Resolution time is down. And Meridian is the first step toward governing the full orchestration system — the AI decisioning layer included — as we scale and fold acquired businesses into the engine.
The operator-builder thesis isn't a slide anymore. It's running in prod.