File / 001 — Portfolio

Noor Sbeih

Status: Open to work — Backend · Full-Stack · Data Eng

Full-stack engineer building systems with real-world stakes — from Modbus-connected factory floors to AI-assisted fitness platforms — and growing a parallel track in data engineering.

READOUT /

View My Work Get In Touch

Fig. 01 — N. Sbeih

About

Location: Palestine — remote-ready
Focus: Full-Stack · Data Engineering
Stack: Django · React · Flutter
Status: Open to opportunities

I'm a full-stack developer specializing in building solutions that bridge web and mobile platforms. My expertise spans React and Django for scalable web applications, Flutter for cross-platform mobile development, and AI/ML integration using TensorFlow Lite. I thrive on tackling complex technical challenges, from real-time industrial automation systems to accessible educational platforms.

I've built diverse systems including industrial PLC control platforms with real-time monitoring, AI-powered fitness applications with personalized recommendations, educational apps for children with hearing disabilities, and multi-tenant management systems for enterprises — and I'm now extending that into data pipelines and analytics.

Resume GitHub LinkedIn

Projects Completed

Years Experience

Happy Clients

Code Commits

Capability Index

05 categories

Frontend

React
Next.js
TypeScript
JavaScript
Tailwind CSS
HTML5
CSS3

Backend

Django
Node.js
Express
Python
PostgreSQL
MySQL
MongoDB

Mobile

Flutter
React Native
Firebase
Kotlin

Data & ML

pandas
NumPy
Jupyter
TensorFlow
Keras
scikit-learn

Tooling

Git
GitHub
VS Code
Docker
Vercel
Figma

Selected Work

Eight records spanning football analytics, data pipelines, industrial IoT, AI-powered mobile apps, accessible education, and enterprise tools.

No. 01

Corner Routines

A World Cup corner as a short sequence: how it was delivered, where the first contact was, and whether the shot came from the second ball.

Built the 2022 World Cup corners from StatsBomb events and 360 freeze frames into sequences in DuckDB. A short corner stays a ground pass near the flag. Inswing, outswing, and straight are the deliveries. The sequence stops at a goal, the next restart, or eight seconds after the first contact. The static Next.js site ranks teams only when they have enough corners, hides a routine percentage when the cell is too small, and plots each delivery end and first contact on the attacking end of the pitch. Prefect runs the build, pytest locks the zones and the second ball, and GitHub Actions runs the Python and web checks.

Python
DuckDB
SQL
Prefect
Next.js
TypeScript
pytest
GitHub Actions

Code Live

No. 02

Football Gegenpress

The eight seconds after a World Cup team loses the ball: who presses, who wins it back, and where.

Built the 2022 World Cup from StatsBomb open data into possession spells, live losses, and eight-second windows in DuckDB. Restarts never start the clock. Pressure and regaining the ball are separate counts, checked against StatsBomb's own counterpress tag. The static Next.js site ranks teams on high losses — the ball given away in the opponent's third — and shows when those regains arrived, so a five-second cut and an eight-second cut can be compared. Prefect runs the build, pytest locks the exclusions, and GitHub Actions runs the Python and web checks.

Python
DuckDB
SQL
Prefect
Next.js
TypeScript
pytest
GitHub Actions

Code Live

No. 03

City Weather Pipeline

Scheduled weather ingest for eight cities: raw, staging, and curated Postgres, with a live read API and dashboard.

Built a batch data pipeline that pulls daily historical weather from Open-Meteo, stores the original response, validates it, and loads a typed table with safe re-runs. SQL window functions produce 7-day and 30-day averages and temperature anomalies. Prefect orchestrates the flow. A thin FastAPI app and dashboard serve the curated table. Tests mock the HTTP calls, GitHub Actions runs lint and tests, and the API is deployed.

Python
PostgreSQL
Prefect
FastAPI
SQL
Docker
GitHub Actions

Code Live

No. 04

Football Data Explorations

StatsBomb open data as a tested Python package, with a FastAPI layer and a Next.js pitchboard for five famous finals.

Packaged match analysis as an importable Python module instead of a notebook: loaders, shot and pass maps, and a Parquet cache so repeat requests do not refetch StatsBomb. FastAPI serves JSON tables and pitch maps generated on the fly. A Next.js tactics board lets you switch finals, xG maps, shot maps, and a player's completed passes. Tests mock the data source, GitHub Actions runs the Python and web checks, and the UI and API are deployed.

Python
FastAPI
Next.js
Pandas
Docker
GitHub Actions
pytest
TypeScript

Code Live

No. 05

Industrial PLC Control System

A full-stack industrial automation platform with real-time PLC monitoring, mixer control, and production management for manufacturing operations.

Built a comprehensive industrial control system for managing three industrial mixers in a manufacturing facility. Features include real-time PLC communication via Modbus TCP protocol, WebSocket-based live monitoring of equipment status, intelligent mix formula calculations with support for both constant and variable components, automated production card generation with PDF reports, and complete raw material inventory tracking. Implemented simulation mode for testing without physical hardware and optimized batch reading of PLC registers for minimal latency. Overcame challenges in handling complex mix calculations for different material types (powders, liquids, water), implementing reliable real-time data streaming with Django Channels and Redis, and ensuring data consistency across concurrent mixer operations.

React
Django
Django Channels
Pymodbus
WebSockets
Redis
Python
Material-UI
+7

No. 06

Digital Story Time - Personalized Children Learning Platform

A comprehensive educational Flutter application designed for personalized learning with interactive video content, adaptive quizzes, and vocabulary building for children with hearing disabilities across different grade levels and sign languages.

Built a full-featured educational platform specifically designed for children with hearing disabilities, combining video-based learning with interactive assessments and vocabulary building. The app features a multi-role authentication system supporting students, teachers, and administrators, with comprehensive student progress tracking through SQLite database integration. Implemented an advanced video player with milestone tracking and analytics, supporting multiple sign languages (Kenya, Rwanda, and Malawian Sign Language) to serve deaf and hard-of-hearing children across different regions. Created a sophisticated quiz system with three question types (single choice, multiple choice, and essay) with immediate feedback and progress persistence. Features include adaptive filtering by grade level, difficulty, and themes, cross-platform desktop and mobile support with responsive design, and comprehensive progress analytics with video completion tracking. The app includes vocabulary-focused content delivery, student profile management with local data persistence, and a modern UI with custom themes and animations. Overcame challenges in implementing cross-platform video playback with MediaKit, designing a flexible quiz system with real-time feedback, ensuring data consistency across student sessions, and creating responsive layouts that work seamlessly on both mobile and desktop platforms for children with hearing disabilities.

Flutter
Dart
MediaKit
Shared Preferences
Google Fonts
HTTP
UUID
Device Info Plus
+4

No. 07

FitScoop - AI-Powered Personalized Fitness Platform

A comprehensive Flutter fitness application that combines AI-powered body performance analysis with personalized workout generation, community features, and equipment-based exercise recommendations for users across different fitness levels and goals.

Built a full-featured fitness platform that leverages TensorFlow Lite machine learning models to analyze user body performance and generate personalized workout recommendations. The app features a comprehensive user onboarding system that collects detailed body metrics including height, weight, body fat percentage, blood pressure, fitness assessments (sit-ups, broad jump), and fitness goals to create tailored workout plans. Implemented an advanced AI prediction system using a custom-trained TensorFlow Lite model that classifies users into performance levels (0-3) based on their physical metrics and capabilities, enabling adaptive workout recommendations that match their fitness level. Created a robust community platform where users can share workouts, write reviews, follow other users, and discover new fitness routines with real-time activity feeds and social interactions. Features include extensive equipment integration with 17+ different gym equipment types and body part targeting (10 muscle groups), comprehensive workout scheduling and tracking system, Firebase-powered backend with real-time data synchronization, image upload capabilities for profile pictures and workout documentation, and detailed progress analytics. The app supports both metric and imperial unit systems, includes Google Sign-In authentication, and provides personalized goal-based training programs (bodybuilding, strength training, powerlifting) that adapt based on user performance assessments. Overcame challenges in implementing cross-platform TensorFlow Lite integration for real-time body performance prediction, designing a flexible workout generation system that considers user limitations and equipment availability, ensuring data consistency across complex user relationships and workout dependencies, and creating an intuitive UI that seamlessly guides users from assessment to personalized workout execution.

Flutter
Dart
TensorFlow Lite
Firebase Core
Cloud Firestore
Firebase Auth
Firebase Storage
Google Sign-In
+10

No. 08

Multi-Outlet Form Management Platform

A comprehensive full-stack web application that enables restaurant chains to collect, manage, and analyze customer information across multiple outlets through branded forms, role-based admin dashboards, and data export capabilities with secure authentication and audit logging.

Built a complete multi-outlet survey and form management platform designed specifically for restaurant chains like M Caffe and Mozaic Restaurant. The system features outlet-specific branded forms that collect customer information without requiring user registration, storing submissions with outlet identification for segmented data management. Implemented a sophisticated role-based authentication system with Global Admin and Outlet Manager roles, where Global Admins can oversee all outlets and manage users, while Outlet Managers can only access their assigned outlet's data. Created a comprehensive admin dashboard with real-time analytics, submission metrics, and KPI visualizations using Recharts for data presentation. The platform includes advanced submission management with filtering, search, pagination, and export capabilities (CSV/Excel) for data analysis. Features a robust user management system where Global Admins can invite new managers via email with secure token-based invitation acceptance flow. Implemented comprehensive security measures including JWT authentication, Argon2 password hashing, rate limiting, CORS protection, and detailed audit logging for all user actions. The system supports flexible form configurations per outlet, email notifications via SendGrid integration, and mobile-responsive design for both public forms and admin interfaces. Built with a scalable architecture supporting multiple deployment environments from development to production with PostgreSQL database support. Overcame challenges in implementing secure multi-tenant data isolation, creating a flexible JSON-based form field system that allows outlets to customize their forms without code changes, ensuring proper role-based access control across all API endpoints, and designing an intuitive admin interface that scales from single-outlet to multi-outlet management scenarios.

React
TypeScript
Django
Django REST Framework
PostgreSQL
SQLite
JWT Authentication
Styled Components
+11

Get In Touch

I'm always open to new opportunities and collaborations. Reach out through the form or directly.

Direct lines

sbeihnooreddin@gmail.com /in/nooreddin-sbeih /NoorSbeih