Principal Product Designer for high-stakes e-commerce, automotive & AI.

An authentication-data platform built for engineers and the LLM agents they ship with. Founder · Principal Product Designer · Full product

Auth57 is the prior-authorization data layer for U.S. healthcare — surfaced as a developer API, a managed MCP server for LLM agents, and a clinician-facing lookup product. Designed end-to-end: brand, IA, dev experience, identity flows, dashboard, and the in-product console — shipped to production on Vercel + Supabase.

• Product strategy & brand. Defined the core thesis ("your north star for auth data"), the four-product surface (Cafe / Compare / Compliance / API), and the visual system spanning marketing, docs, and console.
• Developer experience. Architected the public API reference, response models, and tier-based access — the front door for engineering teams evaluating the platform.
• MCP server. Designed the Model Context Protocol surface that exposes Auth57 to Claude, Cursor, and other agentic tools — defining tool descriptions, auth scopes, and response shapes.
• Atomic design system. Tokens, typography, motion, dark/light — built in Figma and ported to Tailwind v4. One source of truth across docs, marketing, and the in-product console.

Four products · live marketing pages

AI governance for U.S. equity trading — a workspace that makes every AI decision traceable to its source. Sole UX & Product Designer · Production code · Acquisition target

ArtifactIQ governs how AI agents operate inside regulated financial workflows: every intent, policy enforcement, and approval is auditable. Backed by 9 issued patents, targeting Salesforce / IBM / Microsoft acquisition at a $25M–$60M valuation. I founded the design practice — visual language, interaction model, component library, and five production-ready HTML screens shipped from Claude Code into AWS.

• Human-AI collaboration model. A three-column workspace — watchlist, LLM thread, artifact canvas — anchored by a persistent governance lifecycle bar (Observe → Analyze → Propose → Policy → Approval → Execute). The AI's pipeline is visible at every step. Not a black box.
• Explainability by design. Every AI-generated artifact carries a unique seal ID, links to its source wiki document, and routes through a visible approval chain — so a regulator can trace any enforced rule back to the human prose it derived from.
• Multi-persona governance. Four roles (CRO, Portfolio Manager, Policy Manager, Regulatory Observer), each seeing a different slice of the same governance state with role-appropriate actions and an escalation model that mirrors enterprise authority.
• Design system & handoff. Full token system + five standalone production-ready HTML files structured for one-line AWS deployment — collapsing engineering risk at the acquisition milestone.

Loyalty for the 0 Series EV era — a program that lives both in the app and on the in-vehicle screen. Product Design Consultant · Strategy · Concepts

Researched loyalty programs across competing OEMs and re-imagined how Honda & Acura ownership extends into the vehicle itself. The in-car screen becomes a loyalty surface: the moment your earned miles unlock a new model, the car asks if you're ready to upgrade — and offers to drive you to the dealer.

• 3D NFTs of new models — collectible ownership artifacts that bridge the marketing site, the app, and the in-car screen.
• TestDrive in-car notification — an opt-in prompt that surfaces new-model arrivals on the cluster screen of an existing owner, scheduled around the route and not the road.
• Scan-n-Swap — a frictionless transition system for owners moving into a new model: profile, presets, paired devices, and earned loyalty all carry over with a single scan.
• Why this matters for a marketplace. Loyalty is the same problem as repeat-purchase trust — both run on identity, ownership context, and a clean handoff between transactions.

Feed.me — online food shopping with an AI assistant that knows what you actually want to eat. Product Design Consultant & Advisor · Heuristic eval · Onboarding · Main-screen UX

Feed.me is online grocery and prepared-food shopping with a GenAI assistant in the loop — predicting what you want to eat right now, suggesting it before you have to ask, and turning a 20-tap order into a one-tap "yes." I came in as the design consultant after the founding team had a working v1: simplified the feed, tightened onboarding, fixed the main-screen information architecture, and added a UGC upload path so the model gets smarter every meal.

• Heuristic evaluation. Audited the v1 against Nielsen heuristics + consumer-app conventions. Surfaced the friction points that were costing first-week retention.
• Simplified the feed. Re-prioritized signal over novelty — what to eat now beats what to eat someday. The home screen earns the user's trust on the first scroll.
• Cleaned up the interface. Tightened the main-screen IA so AI suggestions, manual search, and saved orders all coexist without overwhelming a hungry user.
• UGC + the AI flywheel. Designed the photo / rating upload path that becomes the training signal — the more people use it, the better the predictions.
• Strategy & brainstorming. Worked directly with the founders on positioning, roadmap sequencing, and what to build next now that the AI loop was closed.

The Apple Core Buy Flow — the canonical e-commerce checkout pattern, exported to a global partner network. UX Principal · Apple Premium Partner · Design System Lead

Apple Premium Partners orchestrates the Global Preferred Partners program — premium e-commerce checkout deployed across partner platforms worldwide. I led the unification: a single bespoke atomic design system, centered on a meticulously crafted Core Buy Flow, that lets each partner customize while keeping the consistent Apple experience.

• Apple Core Buy Flow. Owned the prototype and high-fidelity design for stakeholder review and developer handoff — the canonical Apple e-commerce checkout pattern, end-to-end.
• Atomic design system. Shipped a Figma-based component system from the ground up, anchored in HIG and an 8-point grid, with custom SVG icons, atomic components, and detailed micro-interactions and component states.
• Engineering as the central hub. Annotated specs and interactive prototypes that doubled as a partner-implementation reference. Worked directly with the Magento dev team to engineer parity across Magento templates while accommodating local stakeholder customization.
• Stakeholder alignment. Utilized Figma's collaborative features to drive consensus among global partners — demonstrating behaviors and interactions through prototypes inside a unified framework.

SODAS — an ML-powered safety system mining 14M+ service-bay records to predict defects and trigger NHTSA recall investigations. Principal Product Designer · 0→1 · Lives-at-risk consequence

SODAS — Safety Office Data Analytics System. Defined product vision and SOW for a modular ML platform mining 14M+ service-bay records and 20M+ warranty claims from 40 data sources. Replaced a rejected "dashboard with dials" concept with a workflow-first operations interface praised as materially lifting analyst productivity.

• 0→1 interaction model. Real-time alert feed, advanced search toolkit, case builder, and an RLHF workflow for a 40–80 person analyst team operating under direct regulatory scrutiny — where a missed signal triggers an NHTSA investigation.
• Cross-functional alignment. Built consensus across an 80-person distributed global team — Product, Engineering, Data Science, Regulatory — across five different companies. No direct authority. Design as the shared source of truth.
• IA at scale. Drill-down from global fleet view to regional and country-level incident patterns and defect clusters; 40+ data sources unified into actionable visualizations optimized for at-a-glance triage.
• Methods that earned the consensus. Design sprints, reverse-design workshops, iterative prototyping, ongoing user contact, and direct engagement with all tiers of management.

FindMatch — match a customer's preferences to live inventory across the dealer network, on the manufacturer site. Principal Product Designer · 5 engagements · 4-week strategy sprint

Five UX engagements with Volkswagen, including a 4-week sprint forecasting VW's 5-year roadmap. The marquee ship was FindMatch — an iPad-first responsive web app on vw.com that matches a shopper's preferences (budget, body, range, features) against live regional inventory. The same problem shape as a vehicle marketplace's discovery layer: turn intent into a short, ranked list of real cars on real lots.

• iPad-first responsive. Showroom and at-home contexts both. The matching surface had to feel native on a salesperson's iPad and on a customer's couch — same component library, different density.
• Live inventory at the front door. The match always resolved to a real VIN at a real dealer with a real price — not a configurator dream-build. Trust comes from being able to walk in tomorrow.
• 5-year roadmap sprint. Compressed 4-week sprint with VW product leadership: scenario planning, jobs-to-be-done, and a roadmap that prioritized the customer-facing surfaces against dealer-facing tooling.
• Why this matters for a marketplace. Vehicle-discovery UX is the muscle that turns 50,000 listings into the right next car. Same job at OEM scale or marketplace scale — and I've shipped both.