My strongest work sits where product strategy, system design, and AI product judgment meet. I use technical fluency to make better product decisions, not to make the story about code.
Most of my work sits where product strategy meets system design. I like the stage where the request is still fuzzy, the operating model is still unsettled, and the product needs someone to narrow the real problem.
That has meant different things in different products: reframing a seller-facing dashboard request into an AI workflow intelligence system, shipping SproutRoute end to end, and designing Sprout Math as a privacy-first offline iOS math app for kids.
A seller-facing workflow intelligence system that reframed a dashboard request into decision support. I identified the real problem, chose the wedge, defined the AI vs deterministic boundaries, and shaped the operating model around trust, latency, cost, and workflow adoption.
An anonymized enterprise AI case study on turning a dashboard request into workflow intelligence. The page explains the wedge, AI boundaries, system design, product decisions, metrics, tradeoffs, and what I would do differently.
A hybrid AI travel planner with verified places, safety rules, deterministic packing, and attraction memory. The web app runs on Railway with Visual Crossing-backed international weather, and the native iOS app uses WeatherKit for forecast display.
An offline-first iOS math app for K-5 kids. Privacy-safe by design: no accounts, no analytics, no data collection. 2,400+ question templates with professional narration across 40 lessons in native SwiftUI.
10+ years in complex marketplace and platform environments, working across executive stakeholders, scaled systems, and organizational complexity.
Fluent in RAG, evaluation, AI boundaries, cost-latency tradeoffs, and the operating-model decisions that separate a good demo from a production system.
Comfortable moving from early framing into prototyping, system design, shipping, and the adoption problems that follow launch.
Reframed a seller-facing reporting request into an AI workflow intelligence system. Chose a hybrid inference strategy that balanced quality, latency, and cost; designed the product path toward embedded workflows; and shaped the system around real account-team decision loops. Also own strategy and design for large-scale seller growth incentives, tying program mechanics to seller growth and P&L impact.
Led strategic research across commercial ecosystems. Findings shaped product roadmaps, information architecture, and senior leadership decisions.
Led a large enterprise platform migration across multiple teams, created a PMO and analytics function from scratch, and delivered against executive-level goals for a scaled operations organization.
Managed customer experience across a multi-product payments portfolio, drove material business impact through self-serve and contact-reduction work, and partnered closely with data engineering on automated reporting.
Drove scaled lead-generation operations, led a CRM migration, and designed operational workflows and KPI reporting for a commercial organization.
These working systems test how evaluation, guardrails, and safety review behave in real release workflows, not just in strategy documents.
N-way evaluation across 7 frontier models through a single OpenRouter key. Caught my own production SproutRoute model failing the safety floor - and surfaced a faster, cheaper candidate that passes.
A grounded RAG system with prompt-injection, PII, toxicity, and policy gates built around release decisions, not just retrieval demos.
A governance and fairness tool for teams that want launch criteria, audit evidence, and scenario-based safety checks.
Email is the fastest path. LinkedIn and GitHub are here if you want the broader context.
Open to Principal/Staff PM and AI Product Lead roles (Seattle or remote), especially where marketplace economics, AI platforms, and 0->1 to scale execution intersect.