Creeper Glory
checking…A cinematic, maximalist Minecraft showcase — production-grade motion design.
- React
- Vite
- Tailwind
- Framer Motion
Frontend that feels alive.
Self-hosted AI & real systems.
AI Engineer · Full-Stack Developer · Limoges, France
Two halves of one engineer. On the left, frontend craft — interfaces that move, breathe, and feel expensive. On the right, real backends — distributed systems, computer vision, and self-hosted machine learning that runs on its own infrastructure. And across both, I build with modern AI the way an engineer should — planned, shipped, monitored, and safeguarded at every step.
Six projects, all live below. Each one is one click from working proof.
Frontend that feels alive. Real-time graphics, motion design, and interfaces built to feel expensive.
A cinematic, maximalist Minecraft showcase — production-grade motion design.
Generative “synthetic skies” — a real-time WebGL shader, not CSS gradients.
A 3D dreamscape you move through — instanced spring-physics, bloom & bokeh.
Real backends — distributed systems, computer vision, and self-hosted machine learning that runs on its own cloud infrastructure, not a laptop.
Text → emotion spectrum from a real transformer running server-side. No AI API.
A job pipeline you can watch — serverless queue, retries, dead-letter, observable end to end.
Drop an image → classification + OCR + palette, every model self-hosted. No Vision API.
The six above are original concepts. These are other kinds of work — each one its own room, opened from here.

Real brands, reimagined — a before → after redesign study.
Taking a real company’s site and rebuilding it with intent. Each one ships in two cuts: a Simple version and a Luxurious version.

Real products that solve real problems — built, shipped, and operated.
Not demos. Each one is a product with real users in mind: a real problem, a full scope, auth, tests, and a production deployment I operate.
I'm Ayoub Alkak — an AI Engineer and Full-Stack Developer based in Limoges, France. I build secure web apps in microservice architectures, and I can run machine-learning inference on my own cloud infrastructure — NLP, computer vision, and distributed pipelines — when self-hosting is the right call.
Just as importantly, I build with modern AI. I treat LLMs and agentic tooling as part of the engineering toolkit — scoping and planning the work, implementing it, monitoring every step, and putting guardrails and safety checks around what ships. Directing AI well is its own discipline, and it's one I take seriously.
I care about the whole stack — from a 60fps WebGL canvas to a fault-tolerant serverless queue — and about shipping things that are both beautiful and real.