Innovative Projects Realized

Modèle de prévision dynamique du comportement des renouvellements de polices d’assurance

Le département de Prévision et Analytique de TD Assurance est actuellement en charge de prédire les ventes et les revenues des différents produits d’assurance sur une base annuelle et mensuelle. Prédire les revenues, et ceci de manière précise, est crucial pour l’entreprise et son bon fonctionnement. Les modèles jusqu’alors utilisés ont étés construits il y a un peu plus de dix ans et ont été implémentés avec le logiciel Excel. Ces modèles sont complexes et requièrent trop de manipulations manuelles, engendrant non seulement des erreurs potentielles mais également une perte de temps. Le but du présent projet est de développer de nouveaux modèles, tout aussi performant, sinon plus, de les automatiser et de les dynamiser, notamment grâce au logiciel SAS. Nous nous concentrons sur les modèles de renouvellements et d’annulation de polices d’assurance afin de généraliser le phénomène à l’ensemble des produits d’assurance. Ces deux modèles sont dynamiques car directement liés l’un à l’autre. L’utilisation de séries chronologiques est ici requise pour construire les dits modèles.

L2M – AI Chat Agent for Proposal Generation and Knowledge Reuse

This project is creating a smart chat assistant that uses Artificial Intelligence (AI) to help construction and engineering companies write accurate proposals, like bids for major jobs. The assistant is trained on the company’s successful past proposals, making it a secure, private expert. When a team uploads a new Request for Proposal (RFP), the assistant instantly finds and customizes the best text from their past work to match the new job’s requirements, helping them avoid repetitive, manual copying and pasting. The huge benefit for the partner organization is that they can save hundreds of hours of staff time on proposal writing, make sure all their proposals are high-quality and consistent, and ultimately win more contracts by speeding up the process and improving the final submission.

Machine Learning developer intern working within cross-functional teams to develop and commercialize AI-powered solutions in the Public Services sector – AltaML

The project comprises internships in a variety of technical and business roles within our organization and within our GovLab program. Within the organization, roles include associate machine learning developer, business development associate, communications associate and finance associate. Within GovLab specifically, roles include associate machine learning developer, associate business solutions consultant, and project delivery associate. The diff erence is mainly that in GovLab, there is a focus on public sector problems, whereas in AltaML overall, we work across sectors.
For the Machine Learning Developer intern listed on this particular application, Seyed Arshia Razavi (Arshia) will be working with the Government of Alberta – Primary and Preventative Health Services team as they look to pilot their use case (‘IHDA Public Health Data Navigator’). Albertans are challenged to know what data is available and the specifics of what they are looking for before starting their search, which results in significant eff ort for finding information and and lack of transparency and accessibility to public health information. The project aims to solve this by creating a conversational assistant that makes structured knowledge more accessible without requiring users to know exact dataset details.
? Value: the value of this project is multi-faceted, significantly benefiting the general public and Health Analysts. For the general public, the value is in improved accessibility and transparency of public health information, making it easier to find and retrieve data without prior knowledge. The benefit will be estimated by tracking site analytics, aiming for an increase in the number of visitors, a longer average session duration, and an increase in positive media mentions. For internal staff , particularly Health Analysts, the project is expected to decrease the staff analyst time spent generating insights from the data, reducing the current manual eff ort required for comparing and generating insights and analysis tools.
? Focus: the focus of the project is to apply machine learning, agentic workflows and natural language processing to change the way the public accesses health information.

L2M –An energy efficient technology for enhancing indoor air quality

This project focuses on commercializing a novel heat-based air disinfection technology. It aims to validate market needs, assess customer segments, and develop a viable business model to bring an energy-efficient purification solution to market, improving public health and making air disinfection affordable.

Designing a culturally inclusive information system for Canadian universities to enhance international student recruitment

The number of Asian students getting their post-secondary education abroad has doubled over the past decade to around 2.5 million. In the same period, the number of agents in the admission business in China has increased ten times to nearly 400. High school students in China find it difficult to make informed decisions independently about application and admission to foreign universities because the western university websites and information sessions are not designed keeping their cultural practices in mind. This research project undertakes an iterative, user-centered design of an online, real time information session and related international student information web pages to make them culturally inclusive of prospective students from China. The benefits from such a system for the students could include higher efficiency and reduced time for information seeking and bypassing of third party agencies. The anticipated benefit to Canadian universities would be facilitation of international student recruitment and increase in the number of international student admissions.

L2M – Vision-Based Upper-Limb Assistive Devices

Upper-limb assistive devices remain hard to use because control often feels unnatural and tiring, which leads to abandonment. The intern’s M.Sc. project proposes a non-invasive, vision-based control layer that links what the user looks at to objects in the real-world 3D environment and predicts natural hand paths and joint actions. This approach is intended to lower mental effort, shorten training, and support embodiment so devices feel more like part of the body. Canadian prosthetic and exoskeleton teams can use the results to choose control features, define clear performance targets, and plan integration with existing hardware and software. Through Lab2Market, industry and clinical stakeholders will be engaged to identify adoption barriers, ethics and workflow needs, and pilot designs, creating a practical path to a validated, commercially viable product in Canada’s digital rehabilitation space.

Intrafibrillar vs surface mineralization in bovine flexor tendon-derived collagen fibrils – an in vitro model of calcific tendinopathy of the human Achilles.

Calcific tendinopathy (CT) is a staged, reactive process where pre-calcific changes are followed by a calcific phase (formative, resting, resorptive) and by post-calcific remodelling in which hydroxyapatite-like deposits form and quite often resorb spontaneously. CT is most prevalent in the neck, foot and hip and within the foot. The Achilles is a well-documented site of CT, yet in all cases clinical imaging cannot resolve where mineral reside relative to collagen fibrils (intrafibrillar vs surface/extrafibrillar) which is essential to understand the impact of CT on tendons’ mechanical properties. To better understand mineral deposition in CT, we propose to develop an in vitro model that interrogates a formative-phase-like, acellular window at 37 °C using tendon-derived collagen fibrils from bovine SDF to determine mineral location and its fibrillar-scale nanomechanical signature under controlled conditions. In addition, we will take advantage of a recently published method to locally destabilize collagen molecular packing within single fibrils to test whether mechanically damage is required to nucleate the formative phase in CT.

Development and Validation of the Next GenerationSimulation Platform for High-Rise Buildings

Recent years have witnessed a boom in the construction of modern high-rise buildings in megacities around the world. It is important to design a high-rise building that can effectively withstand both wind and earthquake loads. Nonetheless, in current practice, the design of high-rise buildings for wind and earthquakes is done independently. Hence, there is an urgent need to develop a set of integrated design guidelines for both wind and earthquake loads. The proposed research is focused on the development of tools and methods for efficient and reliable simulation, which will lead to an analysis platform for the understanding of the response of high-rise buildings, subject to both wind and seismic loads. Kinetica is a leader in the design of tall buildings and the simulation tools developed during this project will create working platforms for Kinetica to advance its competitiveness and understanding of tall building design

THE EUROPEAN COURT OF HUMAN RIGHTS AND TURKEY’S “KURDISH QUESTION”: THE ROLES OF LITIGANTS, HRO’S AND THE COURT IN FRAMING AND NARRATING RIGHTS

This qualitative socio-legal research focuses on the role of litigation at the European Court of Human Rights in framing and legally narrating the relationship between the Turkish state and its Kurdish minority, concentrating on cases having to do with prisons and hunger striking. It will investigate the arguments made by litigants and HROs, and decisions made and reasoning provided by the Court. The three primary expected outcomes of the research are: (1) to provide the necessary background research and a draft socio-legal report on these issues that will inform a subsequent publishable paper on supranational rights-claiming and its implications in the context of Turkish-Kurdish relations; (2) to provide broader context to Professor Bartholomew’s research project on Kurdish hunger striking in Turkey as a means of creating rights and demanding constitutional recognition; and, (3) to develop linkages with scholars in Istanbul whose research is on Turkish-Kurdish relations, the ECHR, and democratic politics.

High-fidelity simulation of multiphase and thermal processes in hydrogen storage and refueling

Hydrogen will play a key role in Canada’s transition toward low-carbon energy due to the possibility of carbon-neutral production, its clean combustion, and high energy content per unit mass. As hydrogen is typically stored at high pressure (700 bar or more), rapid compression during refueling can generate substantial heat, which can cause pressure overshoot, underfilling, and thermal stresses in composite or metal-lined tanks. This project will address the challenge of hydrogen storage through the development of high-fidelity computational fluid dynamics (CFD) simulations. In collaboration with U. Oxford, simulations of high-pressure hydrogen refueling will contribute to improving the efficiency and safety of hydrogen storage systems. Overall, the research directly contributes to Canada’s national hydrogen strategy and addresses broad challenges in hydrogen storage, which are also top priorities for the partnering university and nation (UK).

L2M – Validating Philosoph-AI’s AI Governance-as-a-Service (AI-GaaS) Platform

Project Overview – L2M – Validating Philosoph-AI’s AI Governance-as-a-Service (AI-GaaS) Platform / (Mitacs BSI 2026)

Philosoph-AI Consulting Inc., based in Calgary and Geneva, is an emerging advisory firm specializing in artificial intelligence (AI) governance, ethics, and regulatory alignment. This project, led by Sandrine Berezowski (PhD candidate, University of Alberta) under the academic supervision of Dre. Martine Pellerin, aims to transform Philosoph-AI’s consulting expertise into a scalable digital platform that enables organizations to implement responsible, transparent, and compliant AI systems.

As AI adoption accelerates, most organizations—especially small and medium-sized enterprises—lack structured frameworks to ensure ethical oversight and compliance with emerging standards such as the EU AI Act (2024), ISO/IEC 42001, and NIST AI RMF. The project addresses this gap by developing and validating a modular AI Governance-as-a-Service (AI-GaaS) solution that integrates diagnostics, policy templates, and measurable governance indicators.

Over four months (January–April 2026), the project will:
1. Conduct a market validation study with over 50 organizations across sectors;
2. Develop a viable business and pricing model for commercialization;
3. Design an intellectual-property and data-governance framework; and
4. Deliver a comprehensive commercialization roadmap for the 2026 launch.

Through this collaboration between academia and industry, Philosoph-AI will demonstrate how ethical and inclusive AI governance can be a driver of innovation, competitiveness, and public trust. The project strengthens Canada’s leadership in trustworthy AI by linking research excellence with practical business implementation.

Ultimately, this initiative positions Canada as a global hub for responsible AI governance—where ethics, transparency, and innovation converge to shape the next generation of digital decision-making.

GausLight Splatting

Le projet GausLight Splatting vise à mettre au point une nouvelle façon de transformer les données recueillies par des capteurs laser 3D (appelés LiDAR) en modèles numériques réalistes et faciles à explorer. Ces capteurs produisent des nuages de points, un peu comme des milliards de petites coordonnées dans l’espace, qui sont très précis mais difficiles à utiliser directement. Le projet propose d’adapter une technologie récente, appelée Gaussian Splatting, afin de convertir ces nuages de points en images 3D continues, interactives et rapides à visualiser sur ordinateur ou dans un navigateur web. Pour l’organisation partenaire, cette innovation permettra de gagner du temps et de la qualité dans le traitement de ses données, en facilitant la création d’expériences virtuelles immersives destinées aux musées, aux événements et à la mise en valeur du patrimoine. À terme, cette recherche contribuera à accroître la compétitivité de l’entreprise tout en ouvrant la voie à de nouveaux usages de la 3D dans les milieux culturels et créatifs.