Déreck Piché
Curriculum Vitae
Academic Background
Research Master's in Computer Science, Mila & University of Montreal.
January 2024 - Today
Supervised by Aaron Courville, with a thesis in simultaneous learning in MARL. Attended Representation Learning by Aaron Courville, Machine Learning for Robotics by Glen Berseth, Reinforcement Learning and Optimal Control by Pierre-Luc Bacon and Probabilistic Graphical Models and Learning by Simon Lacoste-Julien. GPA: 3.93/4.3.
Honors Bachelor's in Computer Science, University of
Montreal
2021 - December 2023
Used the honor path to take more applied math classes (including the 2 graduate courses Stochastic Simulation by Pierre L'Ecuyer and Dynammic Programming by Emma Frejinger), and to initiate myself to machine learning research with a project on Tensor Network Compositions under the supervision of Guillaume Rabusseau, a core academic member of Mila. GPA: 3.746/4.3.
Professional Experience
Official Tutor at The Department of Computer Science and Operations Research
January 2024 - April 2024
Employed by the department of computer science and
operations research of the University of Montreal to
help bachelor students in all their classes.
Deep Learning Engineer at GPR Geophysics
May 2023 - August 2023
Internship where I worked on a large-scale deep learning
geophysics project. I also did computer graphics to
contribute to the creation of synthetic data.
Projects and Achievements
Utero
December 2023 - Today
Utero is a work-in-progress library for adaptive
algorithms written in Julia. It builds on no outside
dependencies and contains its own auto-differentiation
engine.
https://github.com/ScierKnave/Utero
Tensor Network Compositions
2023
Research on the use of tensor network methods for
machine learning with Professor Guillaume Rabusseau,
which is a core academic member of MILA.
https://github.com/ScierKnave/honor_project/blob/master/doc/rapport_projet_honor.pdf
MNIST Sum Competition
2022
Finished in the first position among bachelor's students
in a Kaggle competition for classifying sums of pairs of
handwritten digits using data augmentation and an
ensemble of convolutional neural networks.
Minimalist Chess Engine
2021
Programmed a minimalist chess engine using the MinMax
algorithm (with α-β pruning).
https://github.com/ScierKnave/chess_bot