Principal Product Designer for high-stakes products — automotive, fintech, marketplaces & AI.

Will this drug get approved? Auth57 turns U.S. prior-authorization rules into a developer API, an MCP server for AI agents, and a lookup app for non-clinical ops. Founder · Principal Product Designer · Full product

U.S. health insurance gatekeeps drugs and services with prior authorization (PA) — a tangle of state-, plan-, and drug-specific rules buried in payer PDFs. Auth57 turns that mess into clean data: which drug, which plan, which state, what conditions, with the source bulletin behind every answer. Surfaced as a developer API, a managed MCP server for LLM agents, and a clinician-facing lookup app. Designed end-to-end: brand, IA, dev experience, identity flows, dashboard, 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

auth57.io/demo · GLP-1 lookup

Drop screenshot → /images/auth57-cafe-glp1.jpg

Cafe — GLP-1 lookup

Type a drug. Pick a state. See the live PA verdict + source.

Open ↗

auth57labs.com/compliance

Drop screenshot → /images/auth57-compliance.jpg

Compliance

Regulatory tracking & tamper-evident audit trail.

Open ↗

auth57labs.com/compare

Drop screenshot → /images/auth57-compare.jpg

Compare

Side-by-side payer-policy diff across states & programs.

Open ↗

mcp.auth57.io · MCP server for AI agents

Drop screenshot → /images/auth57-mcp.jpg

MCP

First MCP server for prior-auth data — native to Claude, Cursor, Zed. One config line, zero custom integration.

Open ↗

auth57.io/watch · reviewer triage interface

Watch

RSS → Claude → reviewer → pa_rules. Every payer bulletin sealed as a tamper-evident artifact. Confirm publishes; Override drafts.

Open ↗

Three themes, one build, one set of tokens: a community app for iOS and Android, shipped to testers through TestFlight. Founding designer and builder · Product, design system, and code

I designed Acts in Figma and built it in React Native, with every color, space, and tappable element in one token file. The client flips between three themes inside the shipped build. A grid audit fixed 176 off-grid values, and the first tester's confusion became a four-screen welcome shipped over the air the same day.

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

Drop screenshot → feedme-dishes-feed.jpg

Dishes feed

Drop screenshot → feedme-flavomatch.jpg

FlavoMatch — taste training

Drop screenshot → feedme-geolocation.jpg

Set geolocation

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.

Keepsi — a spatial memory vault for shared photos, documents, and digital goods. Principal Product Designer · End-to-end · Brand · Spatial UX · Mint / Buy / Sell

Keepsi is a spatial-first memory app — your photos, documents, and digital goods float in real space, shareable to family circles with comment threads, invites, and minted ownership. Designed end-to-end from a blank page: brand, IA, spatial layouts, family-sharing flows, and the Mint / Buy / Sell digital-goods economy.

Drop screenshot →keepsi-asset.png

Spatial collection viewer

Drop screenshot →keepsi-aimee.png

Asset detail · comments

Drop screenshot →keepsi-cards.png

Cards grid

Drop screenshot →keepsi-bulk.png

Bulk upload

• Spatial-first IA. Built the canvas-in-room mental model — assets live in space, not in a feed. Comments, invitees, and edit affordances stay glanceable without occluding the artifact.
• Family-circle sharing. Designed the invite, permission, and comment-threading flows tuned for grandparents-to-grandkids — minimal config, maximum forgiveness.
• Mint / Buy / Sell economy. Branded the digital-goods marketplace and tied it back to the same spatial frame so collecting feels native to viewing.
• End-to-end from blank page. Logo, type, motion, IA, layouts, micro-copy — single pair of hands across every surface.

link.me

link.me — agency-grade analytics and product-creation tooling for influencers, athletes, and creators. Product Design Lead · Agency dashboard · Creator profile · Product-creation wizards

link.me is a creator commerce platform built for the people behind the talent — agencies, managers, and the creators themselves. I designed the agency analytics dashboard, the public creator profile surface, and the wizard flows that take creators from blank page to published merch / app / link in minutes.

Drop screenshot →linkme-agency-analytics.png

Agency analytics — team, stats, merch, apps

Drop screenshot →linkme-creator-profile.png

Creator profile — verified, social embeds, recents/shouts

• Agency analytics dashboard. Designed the multi-talent surface — roll-up stats across all platforms (Instagram, TikTok, Threads, YouTube), per-creator drill-down with active / paused states, and merch / app revenue tied back to each profile.
• Product-creation wizards. Built the step-by-step flows that take a creator from idea → published merch, link, or app — with sensible defaults so non-designers ship without help.
• Public creator profile. The audience-facing surface — verified handle, +Followers count, social-platform embeds, link routing, recents / shouts feed. Tuned for athletes (Donald Cerrone, etc.), influencers, and agencies alike.
• Profile editor + content authoring. Side-panel module library (Custom Link, Multi-link, New Merch) with live-preview. Edit on the left, see the public profile update on the right.

Difference Makers — the teacher lesson portal for Istation's K-12 reading platform. Principal Product Designer · North-Star vision · 3-year roadmap · Teacher portal

Istation runs in U.S. classrooms with 4M students and 90K+ teachers. Difference Makers was the 0→1 product line I designed for them — a new teacher-led lesson portal with a setup wizard, curated collections, state-standard alignment, and a TEKS-aware lesson library. The mockup below is the post-wizard splash: where teachers land after they've configured their grade, standard, and subject, and meet the 1,000-lesson library for the first time.

Drop screenshot →dm-portal.png

0-1 mockup · post-wizard splash for the new product

• 0→1 product mockup. Designed the new product line from a blank page — brand mark, lesson library IA, grade/standard/subject filtering, curated-collection cards, and the post-wizard splash teachers land on first.
• North-Star vision + 3-year roadmap. Defined where the product had to be in 36 months and the staged rollout to get there without breaking classrooms in flight.
• Teacher-led, standard-aligned. 1,000+ lessons married to state standards (TEKS, CCSS), grouped into collections teachers can extend with a single Plus-tap.
• Setup wizard → splash. The wizard collects grade / standard / subject; the splash (shown above) is the moment of trust — the lesson library framed by the choices the teacher just made.
• Scale constraints. 4M students, 90K+ teachers, district-IT contracts — every design decision had to clear privacy, accessibility, and rollout-risk gates before it could ship.

The Apple Premium Partners design system — making Apple's e-commerce templates deployable across the global partner network. UX Principal · Apple Premium Partner (agency) · Design System Lead

Apple Premium Partners is the agency program that ships Apple-grade e-commerce to the global premium-partner network. The Core Buy Flow itself was defined by an internal Apple product team — my role was the implementation layer: a bespoke atomic design system and customization model that lets each regional partner deploy the Apple-defined flows with consistent quality while accommodating local stakeholder needs.

• Atomic design system. Built the Figma 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.
• Partner template model. Owned the framework that lets each region adapt Apple-defined flows (Buy, Compare, Cart, Account) without breaking the consistent Apple experience. Local customization without diluting the brand.
• 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 templates while accommodating local stakeholder customization.
• Stakeholder alignment. Drove consensus among global partners via Figma's collaborative features — 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.