
The foundations of Machine Learning and Deep Learning
A practical guide to ML and DL foundations, regularization, hyperparameter tuning, CNNs, RNNs, Transformers, evaluation, and the model lifecycle.
I ship the ML that answers those questions — and I write my way up the theory of deciding, from statistics to reinforcement learning.
What to show: an event recommender inside a campus app. What to say: a chatbot that handles client intake and common legal questions. What comes next: sales and market forecasts for early-stage startups. Different products, same shape — a model, a Python backend, and a decision someone actually acts on.
m101yosef is a version numberRead the middle of the handle: 1.0.1 — not the launch, the
patch right after it. A 1.0.1 only exists because something
real shipped, something broke, and someone fixed it. It's the most
honest version number there is.
Ship. Notice what breaks. Fix it. Ship again. That loop is the whole brand — how I build, how I write. Nothing here is finished — just versioned.

A practical guide to ML and DL foundations, regularization, hyperparameter tuning, CNNs, RNNs, Transformers, evaluation, and the model lifecycle.

A practical walkthrough of value-based and policy-based reinforcement learning, MDPs, and the main algorithms for engineers getting started with RL.
A practical guide to ML and DL foundations, regularization, hyperparameter tuning, CNNs, RNNs, Transformers, evaluation, and the model lifecycle.
Machine Learning
A practical walkthrough of value-based and policy-based reinforcement learning, MDPs, and the main algorithms for engineers getting started with RL.
Reinforcement Learning

Centralized Mansoura University activity-booking concept informed by competitive research and a student survey, built with the IEEE Vectories team.

A graduation-project concept for an AI-driven startup support system — a generative chatbot, an automated data analyser, and a forecasting module — scoped and designed as a team, not shipped as a product.

UX research on registration-form steps and the effort users actually face at each one, challenging the assumption that fewer steps always means less friction.