Thomas Kientz

Below are projects from my studies at ENSAE Paris

Coding competitions - Hackathons

🥇 1st place in the ENSAE Finance society x Ostum Asset Management 2022-2023 Hackathon, focusing on Recurrent Reinforcement Learning for portfolio optimization.
🥇 1st place in the 2021-2022 Solidarity Hackathon (DataForGood & Eleven), where we measured gender inequalities in cinema using AI. See our project here.
🥇 1st place in the AI & sustainability 2021-2022 Hackathon (HEC Paris & Institut Polytechnique de Paris), where we evaluated the carbon emissions of the daily commutes of a company. See our project here.

Projects

On the effect of removing the background of training images in VAE and GAN performance

Advanced Machine Learning project. Removing the background of training images to see if the models train faster, better, and if their latent spaces allowed for more interpretable control of the generated faces. Below are the effects of the first PCA vector in style space for the greyed-out and the original ("raw") model.

Applied Statistics Research Project

Developed a pipeline for the easy implementation of ensemble models in a pipeline to forecast variations in the VIX index. Check it out here.

Convolutional Neural Network Project

Created a Convolutional Neural Network used to translate drawings into LaTeX, with a pygame GUI. Check it out here. (Not a lot of true formulas I could write with the subset of characters I included...)

Ant Colony Simulation

Simulated ant colonies for optimal path-finding in C++. Optimization was my main contribution. Check it out here.

Neural Style Transfer

Developed a Neural Style Transfer algorithm that generates images similar to one image but in the style of another. Check it out here.

Collapse