PRODUCT/0001

BootleG
stuB

A ticket used to get you into a match, then spend the next thirty years in a shoebox. It carried more than a seat number: who you went with, where you stood, what happened that night. A mobile pass gets you through the turnstile. It isn’t much of a keepsake.

Bootleg Stub is a home for your life in football. Find matches near you or in a city you’re visiting, record the games you’ve attended, and watch your history take shape across grounds, teams, competitions, and countries. Add photos, share stubs with friends, and bring earlier matches into the same story as the one you saw last night.

The digital ticket is the keepsake. The bigger picture is your relationship with the game—where you’ve been, who was there, and where you’re going next..

Brand / Product Design / UX / UI / Print / Development (with AI)

bootleg_stub_1

1,200+
leagues covered

616k
matches in the historical archive

Connecting a fan’s past, present, and next match takes more than live scores.

A check-in app is only as good as what it knows. Live fixtures stream from API-Football — 1,200+ leagues, lineups, goalscorers, events every 15 seconds. Everything older, and everything the API doesn't carry, was assembled from scratch. Fast: the historical archive in eight days, the claim engine in one.

615,902 rows · internationals 1872+ · English league 1888+ · European Cup 1955+

The fixture API ships no coordinates, so the stadium layer is home-made: Wikidata and OpenStreetMap merged into 11,000 grounds bundled inside the app at 2.5 MB — the lookup never needs a signal. The maps aren't fetched either: every stadium tile is rendered in headless Chrome from open map data and baked to a CDN.

11,000 stadiums · 1,790 pre-rendered tiles · MapLibre → Puppeteer → WebP

usa_paraguay_2026
mexico_southafrica_2_2026

The app is free and stays free — no subscriptions, no in-app purchases. The revenue lives off the screen, in a print shop that opened alongside the launch: a city poster series set in the app's own editorial typography — Manchester, Milan, Buenos Aires, Tokyo — country prints on archival matte, and t-shirts carrying the same type system.

Three weeks from the first locked screen to the App Store — live June 4, 2026, Android the same month. Designed, written, and built solo, with Claude Code on implementation.

Free for iOS and Android. Keep the Stub.

white
GetItOnGooglePlay_Badge_Web_color_English

PRODUCT/0002

scAm
Atlas 

The travel warning you need is often already out there—buried in a forum, scattered across articles, or disconnected from the place you’re visiting.

ScamAtlas brings that information together as location-specific guidance, with tools for assessing situations that feel off. I built it around the moment that matters: when you need to decide what to do next.

Brand / Product Design / UX / UI / Content Strategy / Development (with AI)

Built for global coverage from day one.

508

cities covered

175

countries

21K+

scam entries in the database

Creative direction meets AI-assisted development.

ScamAtlas was built with a non-traditional workflow: I directed all design, UX, and brand decisions while using Claude Code (AI pair programming) for implementation. No traditional dev team — just a creative director with a clear vision and the tools to execute it.

Every screen started as an HTML prototype, then translated to React Native. The design system was built first and enforced throughout—no "fix it later" debt allowed.

Tools

Each tool exists for a specific moment — the sketchy Airbnb, the QR code at a street market, the Wi-Fi network you're not sure about. ScamAtlas gives you a tool for each one — not a list of tips to scroll through, but something you can use in the moment it matters.

Scam?
Real-time scam identification. Describe what's happening — get instant AI-powered analysis across a 3-stage matching pipeline: local database → global research → AI classification.
Signal
One-tap GPS location sharing to up to 5 trusted contacts via SMS or WhatsApp. No app install required on the receiving end.
Camera Sweep
Hidden camera guided detection using both cameras — front detects infrared LEDs (no IR filter), rear with flashlight catches lens reflections. The user does the sweep; the app guides them.
Wi-Fi Checker
Network security analysis for public Wi-Fi. Evaluates encryption, identifies potential evil twin attacks, and flags risky configurations before travelers connect.
QR Scanner
AI-powered QR code analysis that evaluates destination URLs before opening them. Catches phishing overlays, malicious redirects, and fake payment QR codes.
City Dossiers
Full city intelligence — scam entries by category, neighborhood heatmaps, emergency contacts, and severity ratings. Downloadable for offline access before you land.

Identity

Dark-first was a battery decision — travelers on the move can't afford a bright UI draining their phone. The restraint became the aesthetic: limited color, generous space, and a principle I kept coming back to — restraint is the premium signal.

icon-circle-dark-anim
icon-circle-pro-teal-anim
icon-circle-orange-anim
icon-circle-pro-teal-dark-anim

.

.

.

.

Safe
#2DD4BF

Watch
#FBBF24

Alert
#F97316

Warning
#F87171

Agents

A static database of 21,000+ scams would decay within months. ScamAtlas stays current through a system of AI agents — each with a specific job, running on a schedule. The agents surface the intelligence. I vet what makes it into the product.

Freshness Agent
Tiered refresh pipeline that verifies every scam entry in the database. Top 50 cities refreshed weekly. All 508 cities touched within 90 days. Each entry gets an AI verification check — confirmed, flagged, or updated — and a new timestamp users can see.

Runs daily · 4 AM UTC / ~7KAPI calls/month
"Scam?" Matching Engine
Three-stage pipeline that runs the moment a traveler describes a suspicious situation. Stage 1 searches the local database. Stage 2 searches global research — travel forums, news, advisories. Stage 3 runs AI classification and generates a severity verdict with recovery steps.
 
Real-time · User-triggered
Geocoding & Zone Engine
Every scam entry gets validated coordinates and a zone assignment. Neighborhoods are normalized via AI, geocoded through Google Places, validated with Haversine distance checks, then clustered into zones using 400m proximity grouping. This powers the heatmap.
 
2,448 zones generated / 10,100 geocoded entries / 400m cluster radius
Discovery Agent
When the Scam? engine finds something that's not in the database — a new variant, a location shift, an emerging pattern — it auto-logs it to a staging table. Entries that accumulate multiple hits get flagged for promotion into the main database. The system learns from its own users.
 
AutoScam logging / ≥3 hitsReview threshold / ∞ Growth loop

The result: a product that gets smarter and more current every day it runs — with AI doing the heavy lifting and a creative director making the final call. That's not prompt engineering. That's systems architecture directed by someone who understands what the product is.

A five-second AI video proves you can make a five-second AI video. The leap from that to “AI expert” is doing a lot of work.

With ScamAtlas, the challenge went beyond generating an output. It meant shaping the product, connecting the data, designing the experience, and building systems to gather, organize, and refresh the information—with me reviewing what makes it into the product.
AI helped with the implementation. It didn’t relieve me of the judgment, the decisions, or the responsibility for what shipped.

white

PRODUCT/0003

BootleG
gUiDes

The 2026 World Cup is the biggest ever. 48 teams, 3 countries, 16 cities, and a stack of official travel guides that read like they were written by the tourism board's legal department.

Bootleg Guides is the one that isn't. 50,000 words on where to actually drink, eat, sleep, and survive the tournament — written in the voice of someone who's been to East Rutherford and has opinions about it. Unofficial. Unauthorized. Unfiltered.

Written, designed, and shipped as a native iOS and Android app.

Concept / Design / Writing / Development (with AI)

I found myself with some extra time on my hands when I had a full rupture of my ACL and MCL in the summer of 2025. As someone who's been to every World Cup since 2014, I used it to write a 228-page, 50,000-word travel guide — honest about which stadiums are a nightmare to get to, which neighborhoods to actually eat in, which fan bars are real and which are tourist traps.

I made a book, then watched FIFA mess things up (more than usual) and politicians scare travelers like this was the movie Hostel. Print was the wrong format for a tournament that changes weekly. So I shipped it as a native app on iOS and Android.

I wrote the guide, developed the identity, and built the app and website using AI-assisted development. The editorial voice, the design, and the product decisions are mine.

No client brief. No committee. No one else to blame.

Practical information, a fan’s perspective, and opinions worth putting my name to.

GetItOnGooglePlay_Badge_Web_color_English