Contact

Because the work is checkable and I've been trusted with real projects since February 2026. Every site and app I've shipped is live, and you can talk to the people I built them for. Age is the first thing people bring up and the last thing they mention after we start.

Going from nothing to shipped without a handoff. I do the product thinking, the design, the code, the infrastructure, and the marketing myself, so nothing sits in a queue waiting for someone else. Concretely: Next.js and TypeScript on the web, SwiftUI for iOS, Python for backends and AI pipelines, and self-hosted infrastructure I run and debug myself.

Speed of learning. A year ago I couldn't write a line of code. Since then I've shipped native iOS in Swift, web products in Next.js, a Python library on PyPI, and CLI tools on npm, and I run my own servers. Nobody taught me any of it and there was no team to hand a piece to, so when something broke at 2am it was mine. Whatever the stack is, I close the gap between not knowing it and shipping with it faster than most people expect.

Machine learning, from the fundamentals up rather than the API wrapper version. I've shipped a lot of AI on top of other people's models and I got tired of not knowing what was happening underneath. So I went back to the start: Python and pandas properly, then the actual training loop, train and test splits, overfitting, why a random forest works. The rule I set myself is that I never watch something passively and I never run a line of code I can't explain. I have two small models I trained myself now. One scores copy as strong or weak, the point being to replace a paid API call inside my own products with a model I own that runs free and offline. The other predicts an athlete's daily readiness from wearable signals like HRV, resting heart rate and sleep, currently on synthetic data with real data next. Building them by hand taught me more than a year of calling someone else's API did.

Anything that needs a team of ten and six months of process, I'm the wrong call. I'm not a designer who will agonise over a brand identity for three weeks, and I don't do enterprise compliance work. I'm good at getting a real thing in front of real users fast and improving it from there. If that's not the problem, I'll tell you.

It depends on scope, so I don't publish a price list. Book a blueprint call, describe the problem, and you get a scoped proposal with a fixed number inside 24 hours. If it isn't worth it you say no and it cost you 30 minutes.

A marketing site is usually days, not weeks. Sean Lee's site went from no brief at all to live in four days. Apps and custom systems take longer, and I'll tell you the real number in the proposal instead of the number you want to hear.

No. Everything is custom, written from scratch. That's slower on day one and much faster on month six, when you want a change a template won't let you make.

You own the code and the accounts, and it's built on a normal stack, Next.js, Postgres, TypeScript, so any developer can pick it up. If you want me to keep maintaining it, I will. I'd rather you have the option than the dependency.

Honest answer: for a while I built things I found interesting rather than things people were already paying to solve, and most of them never found users. I know exactly why now, and it was never the code. That lesson is worth more to me than any of the individual products, and it's why I'd rather do unglamorous work on something people already use than launch a tenth thing nobody asked for.

Helpmarq has over 100 users. My open-source dev tools have over 2,000 downloads on npm and Grip is published on PyPI. Client work is the rest of it: live sites and apps people use every day. Some of what I've built is shipped but not marketed yet, and I'd rather say that plainly than dress it up.

Three things. An open-source design system at design.helpmarq.com, 56 hand-built React components that AI agents can install and use directly, one command per component. PadelUp, an iOS coaching app for padel players with AI video analysis, currently in App Review. And an AI trading agent that commits every decision on-chain before the trade executes, so its track record can't be faked. It runs paper trades while I make sure it's safe.

Before I could code, I spent months planning a Google Ads agency. Offer, pricing, systems, the whole plan. I never got a single client because I never did the one thing that mattered: outreach. Preparing felt like progress and messaging strangers felt terrifying, so I chose the comfortable work and called it strategy. The failure followed me into building products, where shipping was comfortable and marketing was scary. The difference now is I can see the pattern while it's happening. When I catch myself starting a new project instead of promoting the last one, I stop and ask what I'm avoiding.

That being 16 and in school isn't a boundary. Everyone around me believes my only job right now is school and that building real things can wait until I have a degree. I think the premise is wrong. Nothing about sitting in a classroom in Greece stops me from shipping products at night, getting trusted by real clients, or learning machine learning from scratch. Hard doesn't mean unreachable, it means it takes longer and most people quit. A year ago people would have called what I do now unrealistic too.

Because I had no cloud budget. My production infrastructure is self-hosted Postgres and an automation server on two secondhand machines in my room, tunneled to my own domain. It's cheaper, and it forced me to actually understand deployment, networking, and what breaks at 2am. Client projects still go on proper hosting. The homelab is where I learn without a bill attached.

Getting deep instead of wide. I've spent a year going broad, one of everything, and the next stretch is the opposite. I want to work on something people already use at scale, where the problem is making it better rather than finding the first user.