Currently: an AI engine of mine is shipping products on its own, 24/7

I build AI
that builds things.

I'm Louie — a finance professional who taught himself to engineer autonomous AI systems. I designed an AI that finds opportunities, writes its own code, tests itself, deploys, and keeps improving — running around the clock with me barely touching it. Below are the live products it (and I) have shipped.

6+
live products shipped
441k+
records served via API
24/7
autonomous build engine
100%
self-tested before deploy
The centerpiece

An AI that runs itself like a one-person startup

Most people use AI to answer questions. I built one that operates — a self-directed engine that lives on a machine, wakes up on a schedule, decides what's worth building, builds it, verifies it against real tests, deploys it to the web, documents what it did, and scraps what isn't working. No human in the loop on the day-to-day.

The autonomous loop, every run
01
Search
Hunt for a real problem worth solving
02
Decide
Score it; pick the highest-value move
03
Build
Write the code, data & UI
04
Verify
Run its own test suites; no slop ships
05
Ship
Deploy live to the edge
06
Reflect
Document, measure, scrap failures
It governs itself with a written constitution, a credential vault of scoped tokens, persistent file-memory, and hard guardrails (it can't spend money, create accounts, or contact people without me). It caught and fixed its own bugs across deploys. It keeps a public build log. The whole thing is the portfolio piece I'm proudest of — a demonstration that an AI can create real, working software with genuine economic and practical value, on its own.
Selected work

Live products, end to end

Data pipelines, APIs, MCP servers for AI agents, real-time front-ends — designed, built, and deployed. Each link is live right now.

🎯

GetMeBetter

Live · autonomous

A growing hub of accurate practice tools that help people genuinely improve at their hobbies — a poker trainer with a verified Monte-Carlo equity engine, a Valorant aim lab scored with Fitts's-Law throughput, a golf leak-finder benchmarked to real handicap data, and more. Built and continuously improved by my autonomous engine, with every tool gated behind its own automated test suite so nothing ships as slop.

Autonomous buildMonte-Carlo simWeb AudioCloudflare PagesTested engines
Open the live hub →
⚙️

Quota — the engine

Live

The public home of the autonomous engine itself: its mission, a ledger of the ventures it's building, and a running diary of what it ships each day. The proof-of-concept that an AI can run production and improvement without day-to-day human initiative.

Autonomous agentSelf-improvingCron-scheduled
Visit quotaledger.com →
📜

License Verification API

Live · API

A normalized REST + MCP API that verifies US contractor licenses across states — 441,000+ cite-stamped records from official public-records sources, with status-change monitoring. Built so AI agents and apps can consume it directly. Integrity-first: never invents a status or date.

REST APIMCP serverData pipeline4 states
See the live API →
🧾

FreelancerTax

Live

CRA-compliant tax tools and templates for Canadian freelancers — expense trackers, HST calculators, and deduction checklists. A finance-domain product built for a real audience, drawing on my accounting background.

FintechTax / complianceSEO content
Visit freelancertax.ca →
📈

CommissionFlow

Live

Commission management for small sales teams — track deals, calculate payouts, and keep reps and finance aligned. A full-stack SaaS product covering the messy real-world workflow that spreadsheets can't.

Full-stack SaaSReactSales ops
Visit commissionflow.dev →
🛰️

AI Model Deprecation API

Live · API

A cite-stamped REST + MCP feed of which AI models are being retired, when, and what to migrate to — across OpenAI, Anthropic, Google, Azure and more. Built for the agents and tools whose code breaks when a model disappears.

MCPChange-detectionDeveloper tooling
See the feed →
How I work

An unusual combination

🧮

Finance brain, builder hands

I come from accounting and finance, not a CS degree. I taught myself to ship full-stack software and AI systems — so I think about correctness, edge cases, and real-world workflows the way a numbers person does.

🔬

Correctness over hype

Every tool I ship has a verified source of truth and automated tests. My poker engine is checked against known equities; my data APIs never fabricate a value. If it can't be proven right, it doesn't ship.

🤖

I build systems, not scripts

The interesting work isn't a one-off prompt — it's an AI that operates: schedules itself, governs itself with guardrails, remembers, tests, deploys, and improves. I design the machine that does the work.

Let's build something.

I'm most interested in roles and projects pushing the frontier of autonomous AI systems. If you're building agents that actually do things, let's talk.

Built and deployed autonomously — this very page is one of the things the system ships.