Logan Labossiere

Intro

Hey!
My name's Logan and I'm a U3 Computer Science / Artificial Intelligence major at McGill University in Montreal, QC. I'm interested in the theoretical and practical aspects of both AI and Software Development, and I'm always looking for new opportunities to learn and grow as a developer and researcher. I'm currently seeking internships or new grad roles as a software developer, in addition to supervision for a Masters or PhD in AI. I'm also open to collaborations on research or software projects, so please feel free to reach out!

When I'm not working, some of my hobbies include playing piano, volleyball, and pickleball, powerlifting, bouldering, surfing, listening to music, and watching movies.

Whether for music/movie recommendations, to chat about the future of AI, to discuss potential collaborations, or for any other reason, please feel free to contact me using any of the links below! For more information about my academic and professional experience, check out the other sections of my website, or my resume attached below.

• Resume
• GitHub
• LinkedIn
• Email
• Bluesky
• Instagram

Projects

Playlist Synchronizer / Jan. 2025
In collaboration with my friend Eric, I have developed a mobile app allowing users to create playlists shared with their friends across different music streaming platforms. I have mainly worked on the cloud services, specifically the user authentication workflow, the integration with Spotify and Apple Music APIs, the use of webhooks, the optimization of AWS Lambda function calls, the database schema, and the system design and architecture.

Tabulator / Sept. 2023
Tabulator is a Google Chrome extension allowing users to instantly group and delete excess tabs. It was built using JavaScript, HTML, and CSS, and is available for download on the Chrome Web Store at the following link. The app received a CWS Featured badge, indicating that it follows CWS technical best practices and meets their high standards for user experience and design.

Experience

Researcher
COSMO Stochastic Mine Planning Laboratory / 2024-2025
As an undergraduate researcher, I've developed a Multi-Agent framework for stochastic mine planning optimization, integrating Reinforcement Learning, Game Theory, and Combinatorial Optimization for the task of supply chain management under uncertainty. I developed a custom environment, and have conducted experiments showcasing the speed, accuracy, and resilience of the model.

Machine Learning Developer
McGill Robotics Club Drone Software Team / 2024-2025
As a member of our Robotics Team, I've had the opportunity to work on a variety of tasks, primarily the development and refining of a custom real-time object detection model and its integration with our pathfinding algorithm. I've had the chance to work with a diverse group of students, and learned about various aspects of collaborative software development, especially regarding version control systems, code review, whiteboarding, and agile development practices.

Mentor
HighTechU / 2019-2021
During my tenure at HighTechU, a software development academy for high school students, I had the opportunity to mentor students in the area of web development. I assisted students in the ideation and development of their full-stack web application projects over the course of eight weeks, providing feedback from both technical and business perspectives. During our weekly meetings, I ensured adherence to the Agile design process, and I provided assistance to students struggling with syntax and version control software. I also consulted monthly with the program's directors to provide feedback on the curriculum, develop new programs, improve outreach in the community, and plan educational and social events.

Research

A Decentralized Approach to Stochastic Mine Planning Optimization
L. Labossiere, Y. Yaakoubi, R. Dimitrakopoulos / 2024-2025
As an undergraduate researcher at McGill University's COSMO Stochsatic Mine Planning Laboratory, I've developed a Multi-Agent framework for stochastic mine planning optimization, integrating Reinforcement Learning, Game Theory, and Combinatorial Optimization for the task of supply chain management under uncertainty. I developed a custom environment, and have conducted experiments showcasing the speed, accuracy, and resilience of the model. I am currently working on writing and revising the paper for both conference and journal submission.
I presented a less mature version of the project, Decentralized Smart Mining Complexes: Balancing Short-Term Efficiency with Long-Term Targets Using Multi-Agent Reinforcement Learning, at COSMO Technical Day (June 2024, Montreal Canada) and COSMO Travelling Technical Day (Dec. 2024, Perth Australia). I delivered a poster presentation with the same name at the Montreal AI Symposium, and submitted a poster entitled Decentralized Stochastic Optimization of Mining Complexes Using Multi-Agent Reinforcement Learning to the OR/MS Tomorrow Mini-Poster Competition, where I was awarded 2nd place in the Undergraduate category, and was the only Canadian student to receive an award.

Jointly Training Task-Specific Encoders and Downstream Models on Heterogeneous Multiplex Graphs
L. Labossiere, K. Al Handawi / 2024
Dr. Khalil Al Handawi and I developed a novel method for training task-specific encoders and downstream models concurrently, introducing the framework Simultaneous Encoder-Classifier Optimization (SECO). We conducted experiments on both a real-world travel dataset we introduced, entitled "Travel Dubai", and the standard graph benchmark dataset "Amazon Reviews'23". We found that SECO, when combined with state-of-the-art architectures, improves classification performance (F1-score, AUC ROC, AUC PR-RC, and Accuracy), prediction performance (Macro F1), and embedding expressiveness (Silhouette score) by significant margins, while reducing training time relative to approaches using contrastive learning. The paper is currently under review at ICLR 2025.

Education in the Post-ChatGPT Era: Ensuring Fairness and Transparency in AI Mitigation Strategies
T. Fortoul, L. Labossiere (equal contribution) / 2024-2025
In collaboration with my friend and classmate Thomas, I've written a paper discussing the impacts of the newfound rise in unauthorized LLM usage by students. We specifically look at several distinct evaluation methods, including in-person testing, remote proctoring, LLM detection software, and project- and participation- based assessments. We conducted a literature review on state-of-the-art pedagogical and technological literature, discussing both the ethical and practical implications of each method's implementation. We are currently revising the paper for conference submission.

Education

Elements

i = 0;

while (!deck.isInOrder()) {
    print 'Iteration ' + i;
    deck.shuffle();
    i++;
}

print 'It took ' + i + ' iterations to sort the deck.';