The mobile app, the dashboard, the API, the database, and the ML model — all five layers, one pair of hands, for real clients who use it every day.
A consumer product, a national platform, a legacy rescue, and deployed AI — each built from the first layer to the last.
A two-sided home-services marketplace
Connects homeowners with tradesmen across Algeria: bookings, real-time chat, ratings, all 58 wilayas. I built the mobile app on a full clean architecture plus a role-based Next.js admin.
Algeria's circular-economy platform
Citizens, factories, transporters and warehouses on one platform to recycle materials: digital wallet, subscriptions, QR, offline-first. My most complex architecture — 13 independent feature modules, each with data / domain / presentation layers.
Bridging a legacy ERP to the modern web
An Algerian SME runs sales on a legacy desktop ERP over Firebird. Across four generations of solutions I built a sync pipeline pulling ERP data into a live online store — without the legacy system stopping for a single day.
A medical ML system — from model to pocket
Three pieces: a mobile app collecting symptoms, a desktop app managing training data and retraining, and a FastAPI server hosting a live XGBoost model — with answers adapted to the role: doctor or patient.
This is the shape every project I build takes. Scroll and watch the layers light up one by one — with what I've actually shipped at each.
Where the user touches the system. Flutter apps on clean architecture with DI, RTL-first Arabic, offline when it matters — and native Kotlin when it has to be.
Role-based dashboards, citizen portals, and 3D sites. Next.js and React with TypeScript — one frontend sharing the same domain logic as the app.
The API that ties it together: Django REST for heavy systems, FastAPI for models, Python Cloud Functions for events, Hono when light is enough. JWT auth and role separation every time.
Six engines in production — from PostgreSQL and Firebase to a legacy Firebird that cannot be stopped. I write the access layer so the system can change databases without the app knowing.
Applied models, not notebooks: XGBoost deployed behind a live API, DenseNet for a diabetic-retinopathy thesis, YOLOv8 for banknote detection, TFLite on-device.
Associations, centers and companies who use what I built every day — in their words.
Client testimonial slot
Client testimonial slot
Client testimonial slot
Client testimonial slot
Client testimonial slot
Client testimonial slot
Client testimonial slot
Client testimonial slot
Client testimonial slot
Client testimonial slot
↑ empty slots — fill them in src/content/testimonials.ts
Systems for real associations and centers, ML research, native apps, and Figma prototypes — across five different runtimes.
I'm Assem Alia, from El Oued, Algeria. A Master's student in AI & Data Science and a former computer-science teacher. Since 2023 I've built complete systems for real charities, centers and companies — most of them Arabic-first products used by real people every day.
What sets me apart isn't a language — it's that I can carry the whole product: from the Figma prototype, to the app, to the server and database, to the deployed model — and explain it to non-technical people.
Got a product idea, a legacy system that needs rescuing, or an AI project? Write to me — I reply fast.