10+ years across analytics, ML, and product. Currently building AI-native CRM for banks at BusinessNext — agents that have to survive audit, not just demo well.
Each one opens up. The detail underneath is the part that actually matters.
Mechanical engineer → analyst → AI PM. Each step changed the next.
American Express, 2020–2025. Manager · Product Analytics.
The win wasn't any single test — it was the system that let us run them all without losing accuracy.
End-to-end ownership: hypothesis → power analysis → variant design → ship → measure → roll-out.
The system mattered more than any single test.
American Express, as Manager · Product Analytics.
In analytics, the team is the product. You don't ship without them.
AI-native CRM for banks. Every agent output has to pass governance before a human ever sees it.
Banks don't reject AI because it's inaccurate. They reject it because nobody can explain what it did. So the product is built around the guardrail layer: role-based access and masking before the prompt, domain routing in the middle, controlled output and a full audit trail on the way out.
No screenshots of dashboards. The decision I made and what it cost.
Thirteen years, five employers, one throughline: get the system to make a better decision than it made yesterday.
What actually shipped, what broke, and the number attached to it.
Measured, not claimed. Where compression actually pays and where it costs you accuracy.
A demo that went sideways, and the design rule I took out of it.
Agentic systems, eval infrastructure, MCP, and what e-commerce analytics looks like after dashboards.
Open to conversations about AI product roles, agentic systems in regulated industries, campus workshops, and mentorship.