Projets novateurs réalisés

Integrated Digital Transformation and AI Commercialization Strategy for Indigenous Community Development

This project focuses on advancing digital transformation and AI-enabled commercialization strategies to support Indigenous community development in partnership with Lac des Mille Lacs First Nation. The project will develop structured digital systems to organize corporate memory, improve data accessibility, and enhance operational efficiency through the application of AI tools for document management, reporting, and workflow optimization. In parallel, the project will support the commercialization strategy of an AI-enabled energy management platform (Moose AI), including market analysis, business model development, and go-to-market planning tailored to Indigenous and remote community contexts. The initiative integrates technical system development with strategic business planning, while emphasizing community engagement, data sovereignty, and sustainable economic development. Outcomes will include deployable digital infrastructure, commercialization frameworks, and applied case-based insights to support broader adoption across Indigenous communities.

Computational study of the effect of electron-ion coupling on diffusion of defects in silicon

Silicon is the most widely used material in electronics and is now becoming important for emerging quantum technologies. Many of the quantum–relevant “color centers” in silicon form and evolve through the movement of tiny defects inside the crystal. These defects play a major role in how silicon devices behave, but their motion is still not fully understood under the conditions in which real devices are manufactured. During processes such as ion implantation, laser processing, or high–field operation, the electrons in silicon can become much hotter than the atoms. This creates non–equilibrium conditions where electrons transfer energy to the lattice, potentially changing how defects move.

This project will use computer simulations to study how these non–equilibrium effects influence the motion of defects in silicon. By comparing traditional simulations with simulations that include electron–phonon energy exchange, the research will determine whether these effects speed up, slow down, or redirect defect motion. The results will help improve our understanding of how defects behave during device fabrication and may support the development of silicon–based quantum photonic technologies.

EditLikeMe : Apprentissage du style d’édition personnalisé par IA

Trebble développe une plateforme d’édition audio et vidéo propulsée par l’intelligence artificielle, conçue pour automatiser les tâches d’édition pour les créateurs non-professionnels.
Aujourd’hui, notre système d’édition assistée par IA (DeepCut) applique des règles générales d’optimisation (suppression des silences, hésitations, répétitions), mais ne tient pas compte des préférences individuelles des utilisateurs. Or, chaque créateur développe progressivement un style propre : rythme de montage, niveau d’agressivité des coupes, utilisation de certains formats, tolérance aux hésitations, etc.
L’objectif de ce stage est de développer un système d’IA basé sur un modèle de langage (LLM) fine-tuné, capable d’apprendre automatiquement le style d’édition d’un utilisateur à partir de son historique de projets sur Trebble, et de reproduire ce style lors de l’édition de nouveaux contenus.
Le projet vise à structurer les données d’édition (contenu brut, versions éditées, transcriptions, logs), à construire un jeu d’entraînement supervisé, puis à fine-tuner un LLM spécialisé pour prédire des actions d’édition personnalisées.
Ce modèle utilisera des représentations latentes du style utilisateur (« embeddings de style ») comme signal de conditionnement, afin d’adapter ses décisions aux préférences individuelles.
Ce système permettra de passer d’une IA “générique” à une IA véritablement personnalisée, capable d’agir comme un assistant intelligent qui édite selon les préférences implicites de chaque créateur.
Le projet s’appuie sur des travaux récents en fine-tuning de modèles de langage, en apprentissage par démonstration et en modélisation de style, et exploite un volume important de données internes légalement disponibles.

Testing for mito-nuclear incompatibilities in two pairs of hybridizing fish species

This project explores how hybrid fish species use and produce energy by studying interactions between their mitochondrial and nuclear genes. By comparing two fish systems—Cobitis loaches from Europe and Fundulus killifish from North America—the project will identify genetic factors that influence metabolism, stress tolerance, and adaptation to changing environments. These insights are important for understanding how fish respond to climate change and may support more sustainable fisheries and aquaculture practices. The collaboration between the Czech Academy of Sciences and the Université de Moncton brings together complementary expertise in physiology, genomics, and evolutionary biology, creating an enriched international research environment. Both institutions will benefit from shared data, methods, and training opportunities that strengthen long-term scientific partnerships.

Documentation de la démarche régénérative du Lac Osisko

Ce projet vise à documenter et structurer la démarche de régénération du Lac Osisko, à Rouyn-Noranda, portée par le Collectif Territoire. Le lac, marqué par un important héritage industriel, fait aujourd’hui l’objet d’une initiative collaborative réunissant citoyens, chercheurs, artistes, partenaires institutionnels et industriels et la Ville de Rouyn-Noranda. Leur objectif est de réhabiliter le milieu naturel tout en renforçant les liens sociaux et en rehaussant la valeur culturelle et collective du site. Au-delà des actions environnementales, cette démarche cherche à transformer la relation entre la population et son territoire. Toutefois, pour que cette expérience puisse inspirer d’autres collectivités confrontées à des défis similaires, il est nécessaire de mieux structurer, analyser et rendre visible le travail accompli.

Le projet consistera donc à documenter les étapes clés de cette initiative, à analyser ses retombées sociales et territoriales, et à comparer l’expérience du Lac Osisko à d’autres projets de régénération au Québec et à l’international. Les résultats prendront la forme d’un rapport détaillé et d’un outil de synthèse accessible, permettant au Collectif Territoire de mieux communiquer son approche et d’en favoriser la reproduction ailleurs au Canada.

En contribuant à transformer une initiative locale en modèle transférable, ce projet soutient le développement de solutions concrètes pour la revitalisation durable de milieux urbains marqués par un passé industriel.

LLM-Based Agents for Healthcare Vendor Intelligence Extraction

Healthcare organizations in the United States face a complex and time-consuming process when selecting new technology vendors. Important information about healthcare software—such as which hospitals already use it, what electronic health record systems it integrates with, and whether it meets security standards—is scattered across many sources online. This project will explore how advanced artificial intelligence systems can automatically gather and organize this information to make it easier to evaluate healthcare technology companies. Working with Toboggan Labs and its client partner, the research will focus on improving AI tools that extract reliable vendor data from public sources and on designing practical methods to measure the accuracy and cost of these systems at scale. The results will help our client partner maintain a comprehensive and up-to-date database of healthcare technology vendors, strengthening its marketplace that connects hospitals with appropriate solutions. Ultimately, this work aims to reduce the research burden for healthcare organizations and support faster, more informed technology purchasing decisions.

Design and implementation of a data quality framework

Data quality is a foundational requirement to ensure reliable analytical outputs and sustained trust in data products at National Bank of Canada. The project will focus and the design and implementation of a data quality framework within the company cloud infrastructure.

Woman Healing For Change

Woman Healing For Change, a non-profit organization focusing on healing and trauma-informed practices to empower women and children is undertaking a new project to address noticeable challenges such as increasing poverty, homelessness and social exclusion prevalent in the North End Winnipeg. By utilizing the co-operative business model along with mutual aid, the organization hopes to create opportunities for community members to access essential social services, resources and available support to enhance their daily lives. The project will last about four months and have one special assistant. The special assistant will be responsible for coordinating project activities, helping in community engagement, preparing relevant materials and supporting administrative functions. The special assistant will make the business processes more organized and efficient so that the project can yield the intended results.

Crabtree Student Ambassador Digital Literacy Program

The Crabtree Community Digital Ambassador program is designed to connect students with their local communities, empowering them to improve and deliver digital literacy and digital skills training to businesses, stakeholders, and citizens to drive economic growth and development.
The program is aimed to push economic development within rural communities, promote digital literacy, and boost business development by enhancing digital adoption, empowering digital skill use within the community, and working with the community to determine the needs, training, and technology required to improve efficiency and productivity within businesses and citizens. As young people from the community, Crabtree Digital Ambassadors will work with local organizations such as libraries, community centres, and economic development agencies to offer arange of services to community members

L2M-L6 ML Insights & Metrics

More than 90% of pharmaceuticals fail to impact/fail to pass regulatory approval due to toxicity or lack of effectiveness. 50% of those failures have the potential to impact patient lives and existing therapeutics can be repurposed to expand therapeutic options if they can be delivered in smaller dosages to where they are needed most. The project aims to accelerate the development of modified nanoparticles to deliver therapeutics to where
they are needed most (targeted therapeutic delivery).

The technology uses AI to extract and index the qualitative and quantitative data about how nanoparticles are made and how they interact or impact cells and in animal models once administered. The data presented in an analytics platform and search engine, allows users to perform head-to-head comparisons of their formulations with peer reviewed literature. This allows users to have confidence in their work and gain insights on how to design or make nanoparticles suitable for their therapeutic of interest and application.

If successful, this solution will generate enough data to train AI models that can simplify the R&D process by predicting how to make nanoparticles for a specific purpose; to load and deliver a therapeutic to where it is needed most by accounting for safety and how it will interact in the body. This will provide Canada the tools to accelerate the creation of safer and more effective solutions for patients – providing more therapeutic options for tailored therapies for patients.

L2M-L6 Miiya

Aivida Inc. is a Winnipeg-based health technology company developing Miiya, an AI-powered Clinical Decision Support System (CDSS) that helps physicians make faster, more informed decisions at the point of care. Miiya works by searching trusted medical knowledge sources, such as UpToDate and Merck Manual, in real time and surfacing relevant, cited recommendations during a patient encounter, so physicians spend less time looking things up and more time focused on the patient in front of them. This project will advance Miiya from its current prototype stage toward a deployable, market-ready product by tackling three interconnected challenges. First, the intern will build the core AI pipeline that connects Miiya to medical knowledge bases, ensuring every recommendation it generates is traceable to a verified source. Second, the intern will implement the security and compliance infrastructure, audit logging, access controls, and data governance, required for Health Canada’s Class II Medical Device approval, which is a mandatory step before any hospital can formally adopt the platform. Third, the intern will design and build a mobile-first version of Miiya through direct, iterative collaboration with physicians at Health Sciences Centre (HSC) Winnipeg, ensuring the tool fits naturally into how clinicians actually work rather than adding to their burden. The expected benefit to Aivida is significant: by the end of the project, the company will have a technically capable, compliance-ready, and physician-validated platform positioned for its first institutional deployment at HSC, unlocking its path to Health Canada approval, enterprise contracts, and commercial scale across Canadian healthcare institutions.

Honouring the Waters: Indigenous-Led Insights on Marine Shipping and Community Sovereignty – Potash marine exportation impacts

The Clear Seas Indigenous Internship Program aims to provide support to First Nations, Métis, and Inuit communities across Canada by helping them build the necessary internal capacity to research and protect their territories through a re-imagined internship program. Clear Seas provides mentorship to the interns and uplifts existing internal capacity within Indigenous communities. The Clear Seas Internship Program is tackling the real-world impacts of marine shipping on Indigenous communities, with significant cultural, social, and economic consequences. While shipping is vital for trade and economic growth, it often disrupts marine ecosystems, interferes with traditional fishing and hunting practices, and undermines Indigenous governance structures. This project’s scope is to be a leader for more holistic, respectful, supportive, responsive, equitable, and inclusive relationships between Indigenous communities and marine ecospheres through co-produced research spaces. The research centres around the marine ecosphere as Clear Seas can leverage their vast network of marine-related industries and partners. The knowledge developed through these projects will inform future Clear Seas research and guide marine-related policy and strategies