I'm a product manager specializing in the infrastructure behind risk, fraud, and Trust & Safety operations. Seven years at Amazon shipping signal architecture, automation, and platform work that served 10M customers, 2M businesses, and 100M+ annual transactions. Today I'm in pre-launch with my first founder venture, JoinPlay, and building agentic AI tools for PMs.
Every piece of work below has AI doing real work: in the product, in how it was built, or both.
Every activity shows two scores: one for the sport, one for the people.
The rec-sports world optimizes logistics and abandons belonging. JoinPlay is a consumer platform pioneering a new category — social sports integration — built for what incumbents ignore. Sole PM & builder: dual-fit matching system, JTBD research across 13 competitors, architecture, and MVP. Tightening core loops before public launch.
Stress-test a PRD before a single line of code.
An AI review system that runs your product spec past five specialist agents — technical feasibility, metric clarity, stakeholder alignment, edge-case coverage, and data privacy — and returns a readiness score with actionable gaps.
The prompts I reach for when I'm shipping with agents.
An interactive library of my favorite prompts for agentic workflows, PRD drafting, research synthesis, and vibe-coded prototyping. Launching soon — preview what's inside.
Building a consumer platform pioneering a new category — social sports integration — that matches users to rec-sports communities on skill and social fit.
Owned the A-to-z Guarantee platform through four product generations, from manual adjudication to ML-driven automation at 100M+ transactions per year.
Inherited a 42-day backlog on a 7-day SLA, with no PM in seat. Cleared it in six weeks without headcount.
Generative and agentic AI are load-bearing in my workflow, not decorative.
I run structured JTBD interviews, then use LLMs as synthesis partners to cluster quotes, find the disagreements no one named out loud, and draft problem framings I can pressure-test against the transcript.
Every spec I write goes through Spec-Check — my own agentic review system — before stakeholders see it. Five specialist agents catch gaps I'd otherwise ship.
I prototype entire flows end-to-end — not Figma, but working HTML, real state, real APIs — so I can user-test and iterate in hours instead of sprints.
For risk & T&S work I still reason about signals by hand, but use agents to generate test cases, stress distributions, and edge-case catalogs that would take days to assemble manually.
Open to senior IC or founding PM roles in platform, T&S, risk, or applied AI. Los Angeles or remote. Let's talk.