Here a small description of the projects is provided as well as (when applicable) links to the code. In the github repositories you should easily find the pdf reports as well.
Master Thesis: Geometric Deep Learning for CFD (Deep Learning)
Study the use of recent advances in Deep Learning for the prediction of Computational Fluid Dynamics predictions, on both toy datasets and industrial ones (more than 1e6 vertices per sample).
Project done during Spring 2019 for my master thesis with the Computer Vision Laboratory at EPFL under the supervision of Prof. Fua (EPFL), Prof. Formaggia (Polimi), Michael Defferrard (EPFL, LTS2) and PhD. Baque (EPFL).
StaySafeGeneva (Data Analysis)
Project of data analysis that investigate an improvement of google maps in which we also take into account the risk of a given road/itinerary. Project presented at the Applied Machine Learning Days 2018, Lausanne. Project done in group of 3 as part of the Applied Data Analysis course of EPFL, 2017.
GetInVolvo - StartHack 2018
Project of Augmented Reality (developed an app) that gave us the 3rd overall position at the StartHack 2018 in Saint-Gallen. It was merely an application of existing open source library that we combined together. I pitched the project twice (once only with judges and once with judges + jury) Other than me three other persons were part of the project: Ondine Chanon, Martino Milani, Charles Dufour.
Road Segmentation (Machine Learning)
Project of Machine learning consisting of identifying roads from aerial images. Project done in group of 3 as part of the Machine Learning course of EPFL, 2017.
Darcy Flow (Numerical Analysis)
Analysis of the well-posedness and stability of Raviart-Thomas elements by an example of source pollution in a water reservoir modeled by the Darcy equations. Project for Numerical methods for saddle points problems of EPFL, 2018.
EEG classification (Deep Learning)
Project of Deep Learning consisting in classifying EEG signals to predict a left/right movement. Project done in group of 3 as part of the Deep Learning course of EPFL, 2017.
POD HiMod (ongoing in private repository)
Project of Numerical Analysis consisting in investigating a new approach for fluid simulation into veins and arteries. Here the idea is to model a venal bifurcation in which the entering and exiting branches are modeled by HIerarchical MODal reduction and the bifurcation itself by standard finite elements using Principal Orthogonal Decomposition. The code is written in c++ (HiMod) and matlab (POD). Project done for the PACS course of Polimi in collaboration with Andrea Zanoni and using availables libraries such as LifeV and redbkit.
Disclaimer
The projects have often been made as part of groups and as such the credits should go to all groups members. I tried to always include all the names, but in case I forgot, please tell me!