Hi! I’m Dylan Savoia

A Software Engineer with a dedication
to learning and problem-solving.

Here you can find some
of the personal projects I developed during the years.

You can find more about my actual professional experience in my CV.

Key Interests:
AI, MLOps, UI / UX

PPO From Scratch
Using PyTorch

An implementation of
“Proximal Policy Optimization”
using PyTorch.

The code allows to train and run
pre-trained reinforcement learning agents on Gym environments
(now called Gymnasium)

Developed With:
PyTorch 2.1, Gymnasium

A* Visualization
Using SDL2

C implementation of the A* algorithm.

The interface is built using the SDL2 library and allows to draw mazes, save and re-load them.

This project explores the optimality of A* providing a clear visualization of how the algorithm uses its heuristic to reach the goal as quickly as possible.

Developed With: C, SDL2

Py-Torch Sequence
Labeling

An LSTM-based architecture to train Sequence Labeling models.

The project shows that even a tiny LSTM with a tiny dataset can be a great low-requirements solution for small use-cases like DIY / home automation projects.

Developed with: PyTorch 2.1

ToC