An enterprise solution for ad-tech monitoring and governance, with built-in remediation for fixes at mass scale. I’m building it at InfoTrust. Its first contract, with a global food & beverage company, was six figures.

Type: Ad-tech monitoring and governance, with built-in remediation at mass scale
First client: A global food & beverage company
First contract: Six figures
My role: Application Engineer, InfoTrust

FIG. 002 — Monitor, govern, remediate. Simplified.

003

CalTrak

iOS & Android app2026~2,500 downloads

A free, private calorie and macro tracker I designed, built and ship on my own. Native on iOS in Swift and SwiftUI, with a native Android port in Kotlin and Compose. Five ways to log a meal, and no ads, accounts or subscriptions; just an optional tip jar.

I build and ship it from Linux: the iOS builds run remotely on a Mac mini over Tailscale.

Users: ~2,500 downloads · ~500 active
Platforms: iOS (Swift, SwiftUI) · Android (Kotlin, Compose)
Logging: Barcode · search (Open Food Facts, FatSecret, USDA) · AI description · photo of a label, menu or plate · custom entry
Backend: Zero-dependency Node proxy in front of FatSecret, USDA and Gemini
Model: Free. No ads, accounts or subscriptions. Optional tip jar.
Timeline: First commit 2026-04-24 → App Store v2.0 2026-05-12 → 4.5.x by Oct 2026

App Store caltrak.app

FIG. 003 — Five ways in, one small proxy out.

004

Trainer’s PC

Mobile & web app2026~227 ms scan · launching soon

A trading-card collection app for Pokémon (English and Japanese) and Riftbound. Point the camera at a card and the GPU pipeline I built identifies it among ~60k products, with a median scan of ~227 ms.

Launching soon; it isn’t public yet. Around 30k lines of code across about six days of commits, self-hosted behind a Cloudflare tunnel.

Status: Launching soon (not yet public)
Scanning: CLIP embeddings → FAISS → ORB geometric verification → OCR
Speed: ~227 ms median scan
Catalogue: ~60k products · Pokémon EN/JP · Riftbound
Stack: Expo / React Native · Node, Hono, Postgres API · web app
Features: Nightly multi-source pricing, value charts, trading between users
Built: ~30k LOC · 2026-09-28 → 10-03
Hosting: Self-hosted behind a Cloudflare tunnel

trainerpc.com (coming soon)

FIG. 004 — Frame to match in about a quarter of a second.

005

Tag Explorer

Chrome extension2024–~50,000 active users

A Chrome extension I built that monitors a page’s network requests and helps debug the tracking tags firing on it.

Built while I was in tag management at InfoTrust, bringing a software engineering background to analytics problems.

Users: ~50,000 active
Purpose: Monitor network requests; debug tracking tags
Built: At InfoTrust, while I was in tag management & analytics (020)
Release: v4.2.1 · updated 2026-09-11 · Chrome Web Store
Store ID: obhceflecohdhjjomleeakkogaiclhoo

Chrome Web Store

FIG. 005 — Request list, one row inspected. Illustrative.

006

InfoTools / InfoGPT

Internal AI toolRetiredVertex AI · Confluence

While I was in tag management at InfoTrust, I built an internal assistant on Vertex AI, connected to Confluence. It summarised pages, emails and videos and answered questions with Google Search grounding. It has since been retired.

Model host: Google Vertex AI
Knowledge: Confluence, with Google Search grounding
Did: Summarised pages, emails and videos; answered questions
Audience: InfoTrust staff
Built: While I was in tag management & analytics at InfoTrust
Status: Deprecated · past work

FIG. 006 — Company knowledge in, grounded answers out.

007

Carnival Cruise Line

Client engagement2022–25Tealium iQ

Tag management engineering in Tealium iQ for Carnival Cruise Line, as part of my analytics work at InfoTrust.

Scope: Tag management
Platform: Tealium iQ
Context: InfoTrust, tag management & analytics engineering

008

Planet Fitness

Client engagement2022–25Thousands of locations

Internal automation I built for Planet Fitness that had to work across thousands of locations.

Scope: Internal automation
Scale: Thousands of locations
Context: InfoTrust, tag management & analytics engineering

009

3D Print Co.

Company, founded2024–$35k peak month

A basement print farm and shop I founded in 2024, selling handmade keychains. Its TikTok has 6,000+ followers and videos with millions of views.

I run it alongside my full-time engineering job, and keep meeting and exceeding expectations there too.

Peak month: ~$35,000 revenue
Units: 40,000+ handmade keychains sold
Capacity: 10+ printers, basement print farm
Audience: 6,000+ TikTok followers; viral videos with millions of views
Role: Founder · founded 2024

Shop Wall dashboard, 010

FIG. 009 — Ten-plus printers, three figures.

010

Print-farm wall dashboard

Operations displayn.d.Printers · cameras · sales

A wall-mounted dashboard I built for my print farm: live status from the Bambu printers, camera feeds through go2rtc, and a running TikTok Shop sales ticker.

Printers: Bambu, live status
Cameras: go2rtc streams
Sales: TikTok Shop ticker
For: 3D Print Co. (009)

FIG. 010 — One wall, three feeds. Wireframe.

011

Home server & edge

InfrastructureOngoingHome first, then cloud

My small projects and services run on a home server and reach the internet through Cloudflare tunnels. When reliability matters, they move to cloud hosting. Cloudflare also handles Workers and DNS.

Default: Home server
Ingress: Cloudflare tunnels
Edge: Cloudflare Workers · DNS
Escalation: Cloud hosting when reliability matters
Example: Trainer’s PC (004)

FIG. 011 — Start at home, graduate to the cloud.

012

Tailscale mesh

NetworkOngoingDrivable from my phone

A Tailscale mesh joins my laptops, home server, Mac mini and iPhone, so I can drive any piece of work remotely, from my phone included.

Nodes: Laptops · home server · Mac mini · iPhone
Result: All work drivable remotely, phone included

FIG. 012 — Every device reaches every other.

013

iOS builds from Linux

Build pipeline2026Mac mini, remote

I build and ship CalTrak, a native iOS app, from a Linux workstation. The builds run remotely on a Mac mini over Tailscale.

Author on: Omarchy (Arch Linux + Hyprland)
Build on: Mac mini, remotely
Link: Tailscale
Ships: CalTrak (003) to the App Store

FIG. 013 — Native iOS without leaving Linux.

014

Omarchy workstation

WorkstationOngoingArch Linux + Hyprland

My daily driver is Omarchy, an Arch Linux and Hyprland setup. Everything here, iOS included, is built from it.

OS: Arch Linux
Compositor: Hyprland
Distribution: Omarchy

015

AI toolkit

ToolingOngoingT3 Code · Claude Opus 5.5

The harnesses and models behind my method: I design, agents implement, and I review every result.

Harnesses: T3 Code (favourite) · Claude Code · OpenAI Codex
Model: Claude Opus 5.5 (current favourite)
Reach: Driven from any device on my Tailscale mesh (012)

016

Fantasy & NFL modelling

Personal toolsn.d.Modelling tools

Personal modelling tools I built for fantasy football and the NFL.

Domain: Fantasy football · NFL
Status: Personal use

017

Shuffle

Mobile appn.d.Built on the Spotify API

A React Native music-discovery app I built: a feed of music on top of the Spotify API, for finding new tracks quickly.

Stack: React Native
Data: Spotify API
Format: Feed-style discovery

018

MyRender

Side buildn.d.—

A side build, listed for completeness. Ask me for details.

019

Application Engineer, InfoTrust

Role2025–Products & platforms

My move from implementing analytics to building products, centred on the InfoTrust Suite platform.

Period: ~June 2025 – present
Building: 002 InfoTrust Suite
Before this: Tag management & analytics at InfoTrust, from June 2022 (020)
Location: Cincinnati, Ohio

020

Tag management & analytics, InfoTrust

Role2022–25GTM · Tealium · sGTM

I joined InfoTrust in June 2022, right after graduating, to do tag management and analytics engineering for enterprise clients: Google Tag Manager, Tealium iQ, server-side GTM and consent. I brought a software engineering background to analytics problems, and built Consent Mode Inspector, Tag Explorer and InfoTools along the way.

Period: June 2022 – ~June 2025
Built: 001 Consent Mode Inspector · 005 Tag Explorer · 006 InfoTools
Platforms: Google Tag Manager · Tealium iQ · server-side GTM · Consent Mode
Clients: 007 Carnival Cruise Line · 008 Planet Fitness

021

Software Engineer Intern, Cincinnati Incorporated

Internship2021C# → SQL utility

A summer internship in Harrison, Ohio. I built a C# utility that parsed machine output files into SQL.

Period: May – Aug 2021
Location: Harrison, Ohio
Built: C# parser: machine output files → SQL

022

Miami University

Education2018–22B. Software Engineering

My bachelor’s degree in Software Engineering, at Miami University in Oxford, Ohio.

Degree: Bachelor’s, Software Engineering
Period: 2018 – 2022
Location: Oxford, Ohio

023

Collegiate League of Legends

Competitive esports2018–22Diamond

I played competitive collegiate League of Legends alongside my degree, at Diamond rank.

Rank: Diamond
Level: Competitive collegiate

024

First Minecraft server

OriginAge 12Where it started

I ran my first Minecraft server at age 12. The home-server habit (011) started here.

025

Off the clock

PersonalOngoingGolf · home automation

I play: Golf
I tinker: Home automation and home servers
I eat: Tony’s Steakhouse, Cincinnati (favourite)

Fig. 02Bars are years; open ends are still running. 3D Print Co. runs in parallel with my day job. I joined InfoTrust in June 2022 and moved to Application Engineer around June 2025. Every bar opens its entry.

Method

I design the system and direct AI agents to build it. Then I personally QA, security-review and validate every result before it ships.

01
Architect first
System design and architecture are my job. Before any code exists, I decide the shape of the system, its data and its boundaries by hand.
02
Agents implement
AI agents write the code against that design; I direct them through Claude Code, OpenAI Codex and T3 Code. Efficient prompting beats syntax recall.
03
QA every output
I don’t trust anything an agent produces by default. I read, run and check every result before it moves on.
04
Security throughout
I consider security at every step of the build, not bolted on at the end, and every result gets a security review.
05
Validate, then ship
I verify every requirement before anything is published. Quality and speed together; results are the measure.

Prototype · test with real users · iterate or kill

My toolkit

Harness: T3 Code (favourite) · Claude Code · OpenAI Codex
Model: Claude Opus 5.5
Workstation: Omarchy: Arch Linux + Hyprland
Network: Tailscale mesh: laptops, home server, Mac mini, iPhone
Edge: Cloudflare tunnels, Workers, DNS
Hosting: Home server first; cloud when reliability matters

On AI, honestly

Syntax-level recall matters less now. My edge is efficient prompting, systems architecture, and knowing what current AI can and can’t do. Ideas should live or die quickly.

Testimonials

From analytics practitioners who use my Consent Mode Inspector (001).

Nº 1
My data team and I just got back from the InfoTrust Cookieless conference in London and we booked tickets because of how much we used the extension!
George WellingtonDigital Media & Analytics Specialist
Nº 2
This extension goes the extra mile by decoding and displaying gcs and gcd parameters, coupled with explanations that demystify their functionalities.
Marouscha DorenbosAnalytics & CRO Director
Nº 3
This tells you instantly if your website is configured correctly.
Sam CoxFreelance Web Analytics Expert
Nº 4
Perfect for quickly understanding whether a site has consent mode v1 or v2 implemented. Perfect for automating implementation audits.
Anna SinitsynaDigital Analytics Leader