Product work at AWS — GenAI meeting automation, autonomous selling, and the AI-Ready Design System teams use to build in Kiro and Amazon Quick — plus the multicloud rebuild I led at CloudHealth · VMware · Broadcom. AWS details are generalized; the product thinking isn't.
Product work I lead and drive at AWS, and the AI enablement underneath it. Names and internals generalized.
A spec-driven, agent-consumable design system and a paved path from idea to working product. Written for two readers — the people building and the AI agents building with them. Adopted org-wide; AFX Hackathon winner.
A governed GenAI product that automates the manual work around every customer meeting for sellers across AWS's global sales org. Metrics set before the first build, the right model for each workflow, and guardrails from day one.
AI runs the repeatable steps of the enterprise sales motion. The recommendation comes first, the reasoning is visible, and the seller approves. Trust drove adoption; adoption drove results.
An AI co-pilot for the design lifecycle — distills multi-stakeholder research into actionable backlogs, generates wireframes with validation loops, and keeps the design org from re-discovering the same insights every quarter. Pilot feedback synthesized in minutes. Insights routed into Asana as tagged roadmap items. Figma components auto-generated for stakeholder comment — one continuous loop instead of three handoffs between research, design, and engineering.
Org-wide work bringing AI tooling to the broader design organization — a
spec-driven workflow that mirrors Kiro's
requirements.md → design.md → tasks.md,
a shared component library every designer can ship from, and the prototypes
that turn into shipped product faster. Less "designers experimenting with AI",
more "designers shipping AI-native products at the same pace as the rest of
the company."
One project walked end-to-end here: research, decisions, redesign, and the data on what shipped.
