Shamnex is the brand under which Khaldoun builds and ships web and mobile software: marketing sites, landing pages, and product apps such as GoGo Done and the UOS Wayfinder campus guide.

He runs its web properties end to end: the shamnex.com company site with its offer landing pages, the gallery.shamnex.com project showcase, and a clean zero-plugin WordPress rebuild of the company design, on hosting he manages himself.

i4alls is the infrastructure side of Khaldoun's work: several OVH servers he configures and administers himself, combining cPanel/Virtualmin web hosting with a Dokku container platform that deploys his own apps.

The fleet is actively operated: uptime monitoring with e-mail alerting, TLS on public surfaces, routine backups where policy requires them, and written root-cause analyses after real incidents, including load-balancer outages and a full mass-compromise cleanup.

Been There is a travel-tracking web app Khaldoun designed, built, and self-hosts. Users tag cities on a flat world map, drop custom pins, and log trips; the app derives countries and capitals visited, with badges, shareable map cards, and CSV export.

Its content is generated from open data: every UNESCO World Heritage site, 8,800+ airports, plus museums, sights, stadiums, beaches, trails, and festivals across 195 countries. Six interface languages, offline-first storage, and a license-key premium tier, with no account required.

GoGo Done is a home-services platform Khaldoun built as two apps: a customer app and GoGo Done Pro for providers. Requests carry map-pinned addresses, calendar scheduling, and required descriptions; providers get document vetting, an on-my-way status, before/after photos, and a six-digit customer confirmation code that alone closes a job.

Fully bilingual English/Arabic with RTL, OTP login, Firebase push notifications with deep links, and Stripe payments including wallet holds, refunds, and provider payout.

UOS Wayfinder is an independent, unofficial project for the University of Sharjah. It helps students find lecture rooms by their dual code system, check walking time between back-to-back classes, and pick step-free routes; a dedicated badge marks wheelchair-accessible paths. Data is a hand-verified demo set pending university partnership.

Khaldoun designed and built it as an Expo/React Native app with a live web version, release-signed Android builds, a fully RTL Arabic interface, and accessibility testing at 1.5x font scale.

This site is itself a portfolio piece: a bilingual Arabic/English personal site designed and shipped as a Cloudflare Worker on the free tier, with a D1 database, R2 media storage, KV edge caching, and a contact funnel with verified e-mail delivery.

It was delivered through structured phases (baseline, i18n, migration, theme port, QA, cutover) and carries WCAG 2.2 accessibility polish, JSON-LD and BibTeX metadata for academic papers, hreflang language pairing, and sitemap plus observability wiring.

Written Arabic normally omits the diacritical marks that fix how words are pronounced and disambiguated, which makes automatic diacritization essential for text-to-speech, education, and search. The catch is cost: the most accurate neural models are expensive to host and run.

The study benchmarks self-hosted neural networks against cost-effective commercial solutions on standard Arabic datasets, weighing accuracy against real serving cost. Published in the ACR 2026 conference proceedings (Springer, Lecture Notes in Networks and Systems, vol. 1942).

AI image generators promise architects an instant virtual lab for visualizing design ideas. For Islamic architectural heritage, with its geometry, calligraphy, and distinct regional schools, fidelity matters, because plausible-looking but inaccurate imagery can mislead education and restoration work.

The paper evaluates Midjourney's outputs against architectural criteria and defines the major current limits of AI-generated representations of Islamic heritage. Published in the journal Buildings (MDPI) in 2024.

Containers are the backbone of modern cloud applications, and Kubernetes has become the standard platform for running them at scale. Deciding which server should run each workload, the scheduling decision, directly shapes a cluster's performance, cost, and reliability.

The paper surveys the landscape of Kubernetes scheduling algorithms, grouping them by strategy, comparing their goals and trade-offs, and outlining open research directions. Published in the Journal of Cloud Computing (SpringerOpen) in 2023, it has become a widely used reference for researchers and practitioners.