Loading

The stack. What I reach for when I design, build, and ship — and what keeps it safe to run in public.

AI development

Building with models and building alongside them — the assistant on this site runs on this row.

OpenAIResponses + Realtime APIs power this portfolio's assistant, text and live voice
Claude Codeagentic pair for building, refactoring, and reviewing
Cursorday-to-day editor
Prompt engineeringstructured outputs, guardrails, and evals over vibes

Design

Where the taste happens before the code does.

Figmaevery interface starts here
Motionevery animation on this site
Magic UIcomponent craft and community, from the inside
OGL / shadersthe WebGL gradient surfaces across this portfolio
Canvafast social and brand assets

Engineering

The production half of design engineering.

Next.jsthis site, and most shipped work
Reactsince the beginning
TypeScriptnon-negotiable
Tailwind CSSdesign tokens that survive a team
Node.jsAPIs and tooling
PostgreSQLthe default database
Pythonscripts, data work, AI glue
Gowhen a service needs to be small and fast
Dockerreproducible everything

Productivity & shipping

The loop between an idea and a deployed URL, kept short.

GitHubcode, reviews, and shipping in public
Vercelpreview deploys as the review surface
pnpmpackage management without the wait
Cal.comscheduling, including the booking flow on this site

Security

How the AI features on this site stay safe to run in public — practices first, products second.

Cloudflare Turnstilebot defense in front of every paid assistant request
Signed conversation tokenshistory is HMAC-signed server-side; the client is never trusted with it
Rate limits & allowlistsper-session and per-IP budgets, strict origins, one request in flight
Kill switchesevery paid surface can be turned off with one environment flag

Have something you want built?I'd like to hear about it.

Work About Stack Roadmap Writing Contact Ask AI

X GitHub Discord

© 2026 · Prishtina, Kosovo