Mathys Loiselle Portfolio

1. About

I am Mathys Loiselle, a Computer Science Honours student at Concordia University with a minor in Mathematics & Statistics. My research lies at the intersection of machine learning and neuroscience. I am currently an undergraduate researcher at Mila, working on multimodal neural decoding, and I lead the machine learning team for Space Concordia's Space Health division. I plan to pursue graduate studies to bridge insights from machine learning, computational neuroscience, and mathematics.

1.1 Current Affiliations

Concordia University✦Mila - Quebec AI Institute✦Space Concordia

2. CV

Open PDF

2.1 Education

Bachelor of Computer Science, Honours

Concordia University - Montréal, Québec

September 2022 - December 2026 (Expected)

Pursuing a Minor in Mathematics & Statistics alongside my Honours degree. Recognized on the Dean's List from Summer 2023 through Fall 2025, and awarded the Québec Perspective Scholarship ($5000, 2023-2024). Coursework includes Machine Learning, Computer Vision, Linear Algebra, and Multivariable Calculus I & II.

2.2 Experience

Undergraduate Student Researcher

Mila (Quebec AI Institute) - Montréal, Québec

March 2025 - Present (1 year)

Extended the POYO neural decoding framework for multimodal neural decoding under Dr. Guillaume Lajoie and Dr. Matthew Perich,training models with multiple masking strategies for multi-session pretraining and single-session fine-tuning. Ran large-scale multi-GPU experiments on Mila's SLURM cluster with WandB tracking, and presented a poster at the NeurIPS 2025 Workshop on Foundation Models for the Brain and Body (BrainBodyFM) and at the Montréal AI & Neuroscience (MAIN 2025) conference.

Lead Machine Learning Engineer - Space Health

Space Concordia - Montréal, Québec

February 2025 - Present (1 year 1 month)

Leading a team of 10 Bachelor's and Master's students on brain health research for long-duration spaceflight. Designed a transformer-CNN architecture for predicting structural brain changes from longitudinal MRI scans, and built the preprocessing and training infrastructure for multi-timepoint neuroimaging data.

AI Software Developer - Internship

TaylorMade Golf - Carlsbad, California

September - December 2024 (4 months)

Built a full-stack chatbot with both text-to-speech and speech-to-text voice interface using Azure Speech Services (Angular/C#), and developed a color profile matching system to align RGB display output with CMYK printing (Python/React). Also built anomaly detection using Mahalanobis distance to flag invalid product configurations before manufacturing.

Computer Vision Engineer

Nano Stride - Montréal, Québec

October 2023 - January 2024 (4 months)

Led computer vision development for Cody, a 5'8" robot answering campus questions via natural interaction. Designed speaker identification combining timing delay of arrival (TDOA), natural language context, and pose estimation for multi-speaker scenarios, and built real-time head motion control using OpenCV and tiny-YOLO to direct robot gaze toward active speakers.

Dataset Annotator

Concordia Applied AI Institute - Montréal, Québec

November 2023

Performed dataset annotation for ML model training and testing, collaborating with a team to optimize data quality and consistency.

2.3 Technical Skills

Languages

✦Python
✦JavaScript
✦TypeScript
✦Java
✦C#/C++/C

AI/ML

✦PyTorch
✦NumPy
✦Scikit-learn
✦TensorFlow

Web & Cloud

✦React
✦Angular
✦Node.js
✦HTML/CSS
✦Azure
✦Firebase

Tools

✦Git
✦VSCode
✦Jupyter
✦SLURM
✦WandB

2.4 Research Interests

✦Machine Learning
✦Computational Neuroscience
✦Neural Decoding
✦Mathematics
✦Robotics
✦Natural Language Processing

3. Research

Google Scholar

Towards a generalizable, unified framework for multimodal neural decoding

N. Krishna, M. Loiselle, A. Ryoo, M. Perich, G. Lajoie

NeurIPS 2025 Workshop: Foundation Models for the Brain and Body (BrainBodyFM)

4. Projects

ConUHacks X - Degree Requirement Visualizer

January 2026

At ConUHacks X, our team built a React/TypeScript & MySQL web app that uses the Gemini API to extract and visualize any university's degree requirements. The platform generates optimal prerequisite path visualizations and provides career alignment tracking, helping students map out their academic journey with AI-powered course planning.

Website GitHub Demo

ConUHacks IX - Concordia Virtual Tour++

February 2025

At ConUHacks IX, Quebec's largest hackathon, our team developed an innovative solution to revolutionize Concordia University's virtual campus tour experience. The project placed in the top 5 among 200+ submissions, demonstrating both technical excellence and practical utility. We leveraged 3D Gaussian Splatting, a cutting-edge machine learning technique in computer graphics, to transform traditional static images into a fully immersive 3D environment that visitors can freely explore.

Website GitHub Demo

Reinforcement Learning Rocket League Agent

December 2024

A reinforcement learning project that trains an agent to play Rocket League, a physics-based vehicular soccer game. The system implements a PPO (Proximal Policy Optimization) architecture to teach the AI agent complex 3D movement, ball control, positioning, and basic-to-intermediate strategy, effectively translating high-dimensional games states into meaningful actions.

Smart Marshmallow RC Tank

June - August 2024 (3 months)

As part of a multidisciplinary Concordia IEEE team, I helped build an innovative RC car from scratch. The vehicle included a marshmallow launcher guided by an automated targeting system using MediaPipe's pre-trained computer vision model for facial and object detection. The project was showcased at Concordia's Frosh 2024, where it playfully launched marshmallows into students' mouths on command.

Website GitHub

Cody AI Service Robot

October 2023 - January 2024 (4 months)

At Nano Stride, Concordia's leading AI and robotics club at the time, I contributed to the development of Cody, a nearly 6-foot service robot designed to interact with passers-by. I focused on Cody's computer vision system, implementing detection of people, facial expressions, body pose, and other subtle cues. This experience not only deepened my technical skills but also sparked my passion for machine learning and the problem-solving mindset it demands.

5. Volunteering

HackDécouverte

November 2025

Concordia University - Montréal, Québec

✦Ensured smooth event logistics and positive participant experience for CEGEP and high school hackers
✦Provided bilingual support for technical project questions and hackathon logistics