hey, I'm blank.
18, based in Romania. I build software on my phone, no PC involved. Been vibecoding since 2023, and most of what's below came out of that.
what I actually do
I'm blank, 18, from Romania. Technology's been the main hobby for years, but what I spend time on has shifted a lot. There was a stretch where Android was basically the whole thing, rooting phones, flashing custom ROMs, swapping launchers, tweaking every setting that could be tweaked. That's mostly faded now.
These days it's AI and building things. I don't own a PC, my phone is the only device I have, and it's what I use for everything: writing code with AI, GitHub, testing sites, deployments, APIs, managing domains, all of it. It's not a workaround I'm proud of or complaining about, it's just the setup I've got and the one everything here was built on.
Three projects make up most of what I've worked on: OrbitOS, Termio, and VexAI, which is what I'm currently putting most of my time into.
what I've built
Termio
currentTermio is a real terminal in your browser, powered by OpenRouter Shell. It gives you a temporary hosted Linux environment you can actually use. Commands run for real, inside that environment. It's not a fake terminal typing animation and it's not a chatbot dressed up to look like a shell.
You get access to normal Linux tooling, Python, Node.js, npm, the kind of thing you'd expect from an actual box, just spun up temporarily and reachable from a browser tab.
Actually works, it's not a demo. Open it and try running something.
VexAI
in developmentVexAI is what I'm spending most of my time on right now. The idea is to give smaller AI models access to a lot more than a basic chat box: agents, tools and actions, work-style functionality, and the ability to use smaller models for genuinely complex tasks instead of just short replies. It's built to work across multiple AI models and providers through one interface, rather than being tied to a single one.
It's not another AI chatbot. The point is letting models actually do things, not just talk.
The UI isn't built from scratch. It's a direct evolution of the OrbitOS interface, carried over and reworked rather than rebuilt. The migration is still messy in places internally, there are pieces that haven't even been renamed yet, like the database layer still called OrbitDB when it should really be VexAiDB by now.
- AI agents
- tools / actions
- multi-model support
- multi-provider
- unified interface
OrbitOS
archivedOrbitOS was a browser-based operating system project, an attempt to build an OS-like environment that lived entirely inside the browser, no install, no backend server doing the heavy lifting. Over its lifetime it grew a real set of features: apps, browser-based storage, a working file manager, multiple themes, a software update interface, benchmarks, a news section, and some early AI functionality layered in toward the end.
It changed a lot over two years and went through several complete UI generations as ideas about what it should be kept shifting. It's not under active development anymore, but the last real iteration of its interface didn't disappear. It became the starting point for VexAI.
- apps
- file manager
- browser storage
- themes
- update UI
- benchmarks
- news
- early AI features
how the projects connect
Started building a browser-based OS-like environment. Apps, a file manager, storage, themes, all running in the browser.
Several UI generations and feature sets. Update interfaces, benchmarks, a news section, early AI features bolted on along the way.
Not actively developed anymore. It's an archived project at this point, not something getting new features.
Rather than build a new interface from zero, VexAI reuses and evolves the OrbitOS UI.
Currently the main thing I'm working on. Still unfinished.
An estimate, not a promise. Could shift either direction.
vibecoding
I've been vibecoding since 2023. I usually start with an idea, describe what I want, then spend an unreasonable amount of time arguing with an AI until it actually works. Everything gets built, tested and deployed from my phone. Right now I mostly use Google Jules, Gemini, ChatGPT and Mistral AI for coding. Which one I use depends on what I'm working on and which one decides to cooperate that day. I don't really care whether every line was written manually or generated. If I understand what it does, can fix it when it breaks, and the end result works, that's enough for me.
That's it, really. It's just how I build things now.
current setup
what I'm using right now
tools I've used over time
built from a phone
No PC, none planned. Everything on this site and everything in the projects above was built, tested and deployed from a phone. That includes the parts you'd usually expect to need a proper desk for.
- AI coding tools daily driver
- GitHub commits, issues, repos
- Testing sites live, on-device
- Deployments shipping from the phone
- APIs & domains managed, not just used
Android
Before AI took over most of my time, Android modding was the main thing. Rooting, custom ROMs, launchers, general system-level customization, a lot of hours went into that. It's mostly in the past now, my interests moved toward AI and building web projects instead. Still like Android and Samsung, just not the center of what I do anymore.
find me
GitHub is currently empty since most of my repositories are private.
moderator of
Community for Key Launcher, an Android launcher built around a T9-style keypad instead of the usual app grid. You navigate apps, search and shortcuts through the numpad, closer to an old-school dialer than a modern home screen.
Community for Brick 1100, an app that turns a smartphone into a simulation of the Nokia 1100. Monochrome screen, keypad controls, Snake, and the rest of the classic brick-phone experience running inside a regular phone.