here is a repository of short personnal projects that are still fun to implement. I dedicate no more than 1/two days on this kind of project.
The sudoku project consists in an automated sudoku solver able to solve 70 sudokus per second no matter the difficulty and solved 1,000,000 sudokus from a kaggle database in less than 4 hours with a success rate of 100%. This project has been coded from scratch using the numpy library.
The carLearning project consists in a genetic approach to solve the self driving car problem. A demo of a typical generation is also visible in the project. /!\ this video is from the implementation I did in 2020. The simulation runs at roughtly 70 frames per seconds using the pygame library to display it. The collision detection system, the neural network creation the offspring generation, the physics engine and the display of the race are coded from scratch using numpy and pygame libraries.
It would be nice to add a real circuit, and better graphics.