Turn your data from something you look at into something you run on.
Most companies collect plenty of data and still run on gut and argument. Reports describe what happened; they don't decide what to do next. Closing that gap is the job — and it's rarely a tooling problem.
I'm the senior data and AI leader whose mandate runs the whole arc: find where your data can become a product, a decision system, or a new line of business; get the model of your business right — what each number means, where the real decisions are made; then build the systems that capture it. Running on your data is the means. New revenue, recovered revenue, and data-powered lines of business are the point. The model is the hard part; the working system is how you know it was right.
And if you're earlier than that — standing up your first data function, or deciding which AI to trust before you commit — the same structural questions apply, asked before the pain instead of after.
- 20M+ membersData products and decision systems operated at national consumer scale.
- 20K+ merchantsBuilt and led data capabilities for a two-sided loyalty marketplace.
- 15-person teamBlended data organization built and led across data science, analytics, and engineering.
- $50M+ annual revenueProduction data and AI products my team owned outright.
What I do
I’m brought in as senior data and AI leadership with a mandate that runs the full arc — from finding the opportunity to shipping the product that captures it. One through-line holds every engagement together: get the underlying model right first, then build.
Find. Where your data can become a product, a decision system, or a new line of business. Getting the model of your business right — what each number means, where the real decisions are made — is how the real opportunities surface instead of staying buried in reports.
Prove. The smallest engagement that validates it: trusted metric foundations, and AI you’ve evaluated instead of assumed. Part of the job is judgment about where ML and automation earn their cost and where they don’t.
Build. Production systems I own outright, that turn the validated model into something the business runs on — and earns from.
Current build: an on-prem evaluation platform that tests LLM vendors, models, and configurations against an organization’s own data — because "which model should we trust with this?" is an empirical question, not a procurement one.
Why it holds
Once the model underneath is settled and trusted, the work compounds instead of calcifying. The next product you want — the next decision system, the next data-powered line of business — starts from a definition it can rely on, not one it has to route around. The self-serve dashboards and AI you actually wanted start working too, because people trust the numbers they run on.
You can’t rebuild your way out of the wrong foundation.
Proof
Ready when you are.