Sulaymon Ruziboev
Open to junior roles — 2026
Designer turned developer. I build desktop apps and web platforms end to end — database, backend, and the interface people actually touch.
01 — About
I started in motion design, then learned to code to build the tools I wanted. Two years and three working products later, I still build the whole thing myself — database, backend, and the interface people actually touch. Based in Tashkent, Uzbekistan, and looking for my first developer role — remote or on-site.
02 — Selected work
A video editor where every edit is one command in a single registry — the interface fires them, and the AI fires the same ones. Built from scratch: my own Rust core and my own interface, down to the timeline and the render pipeline.
Read the case studyFour more builds — desktop, backend, WebGL and the site where it started.
A Windows desktop app that watches the clipboard: copy a reel link and it downloads the clip, works out which channel it belongs to, and releases it on a per-channel interval so a burst of links doesn't become a burst of uploads.
Source isn't public by default — the site has a request page, and I hand out an access code to anyone with a good reason to read it.
A membership platform for digital artists: work stays public, supporters subscribe to monthly tiers and unlock the studio — 4K files, PSDs, process videos. Accounts, pledges and per-artwork access are enforced in the database with row-level security.
Source
Generated soundscapes, slow drifting visuals, a sleep timer and a night mode that dims when you stop moving. Zero dependencies — every world is a raymarched GLSL shader and every sound is synthesised at runtime from maths.
Source
My graphic and motion design work, and the first website I ever made — stitched together from effects I took apart across the web, then rebuilt once I understood how they worked. Four static pages, a Three.js hero, and 36 projects behind a filterable grid.
Source03 — Capabilities
Multi-page apps on Postgres — accounts, membership tiers and per-item access enforced in the database with row-level security, not trusted to the browser.
Windows apps people install and leave running: native windows, background schedulers, OAuth into third-party accounts, packaged into an installer that just works.
Joining systems never meant to talk to each other — clipboard to YouTube, ffmpeg pipelines, jobs that pick themselves back up after a failure.
Design was my first trade, so interfaces get built from tokens up — and WebGL or scroll animation when a page needs to be remembered, not just read. This page is one of them.
04 — Toolkit
AI writes drafts fast. Deciding which draft survives is still the job — and that part hasn't been automated yet.
Where most of the code starts — an agent with repo-wide reach, kept on a short leash.
Architecture arguments, second opinions, and the debugging you can't grep your way out of.
Reading the docs faster than the docs read, and getting a layout on screen to argue with.
Key visuals, textures and thumbnails — generated, then taken apart and rebuilt by hand.
My first trade. Everything generated still lands here before anyone else sees it.
Footage that never existed and a voice to carry it — b-roll, transitions, narration.
Routing models and wiring services together so the boring half runs without me.
05 — Build log
Graphic and motion design — brand identity, channel art, key visuals, After Effects. My first skill in this industry, and still the reason I notice when something sits four pixels off.
My first website, assembled by taking effects apart from across the internet until I understood why they worked. Nothing taught me faster than breaking someone else's code.
Shorts Pilot. A desktop app watching the clipboard, pacing uploads across channels, holding OAuth tokens safely. The first thing I built that works while I'm asleep.
Galera. Accounts, membership tiers and per-artwork access — enforced by row-level security in the database rather than by trusting the browser.
Cutlass. A video editor where every edit is one invertible command object, so an AI and a human have exactly the same reach. In progress — the hard part is the architecture, not the buttons.