Mohamed A.

Development of a complete web platform for WePropre, a startup specializing in professional cleaning services. This solution allows for comprehensive operational management: assigning staff to tasks, automatic calculation of costs and profits, and customer acquisition through a modern showcase website.

Administration panel includes management of tasks, staff, and customer requests. Intelligent system for managing staff status based on actual task schedules. Responsive interface with animations, contact form, and modern design optimized for conversion.

Technical stack: Python/Flask for the backend, HTML5/CSS3/JavaScript for the frontend, SQLite for the database. Production deployment on PythonAnywhere with automatic timezone management for precise real-time task tracking.

Results: operational platform enabling complete daily activity management, automated financial calculations, and an intuitive interface facilitating decision-making.

As part of this project, I designed and developed a complete image classification pipeline, from raw data to web deployment.

The core of the project lies in the from-scratch implementation of three ML algorithms — linear regression, RBF (Radial Basis Function) network, and MLP (multi-layer perceptron) — developed in Rust for performance reasons, then exposed to Python via an FFI (ctypes) interface.

The pipeline includes a data augmentation step (rotations, brightness adjustments, flips, blur…) multiplying the dataset by 11, rigorous overfitting management with train/test split, serialization of trained models, and deployment via a Flask API enabling real-time classification from a web interface.

Stack: Rust · Python · Flask · PIL · ctypes (FFI) · Jupyter