Mohamed Hamlil
Français GitHub

The whole degree, with the numbers attached

The Licence MIASHS and the M1 Artificial Intelligence at Université Grenoble Alpes, published so that every report renders from the source that produced it.

Four projects that carry a number

ProjectResult
Battleship AI
Java, pair project
Four targeting strategies over 300 games each. A placement-density heatmap wins 74.3% and clears the board in 54.8 shots against 94.3 for random fire, while beating a Monte Carlo simulation that costs far more to run.
Grenoble housing prices
R and Python, group project
Hedonic pricing on DVF open data. The log transform is what makes the linear model work, lifting R² from 0.096 to 0.275; a random forest then cuts error from €265,214 to €58,750 by giving up the interpretable implicit prices the method exists to produce.
Perfume satisfaction
R and Quarto, group project
Binary classification over roughly 24 000 fragrances. Four models compared on ROC and confusion matrices; the random forest leads at AUC 0.693, and every model sits between 0.65 and 0.69, which is the honest finding.
Full-stack web
Node, Express, SQLite, solo
Nine practicals ending in a complete server with persistence, password hashing, sessions and authentication.

Browse it

Site The published site A page per course, charts drawn from the result files, and the Battleship bot playing a recorded game. Repository Source Source next to rendered output, so everything reads without cloning.