Innovative Projects Realized

L2M – Next Generation Data Validation and Annotation Utilizing Geometrical Space of Differentiable Labelers

The challenge to be addressed in this project is to develop a next generation of machine-learning-guided data validation and annotation tools that take account of differences in human judgment and decision-making processes.

Réutilisation efficace des modèles prédictifs à base d’ensemble d’arbres pour faire face à de nouvelles tâches de tarification

En assurance de particuliers, on retrouve plusieurs tâches partageant une majorité de caractéristiques, mais où pourtant chacune d’elles est unique. En assurance automobile par exemple, la tâche de tarification au sein des différentes provinces est unique, étant donné que chacune d’elles possède ses particularités, bien qu’elles se ressemblent beaucoup. De ce fait, il ne peut y avoir un seul modèle pour résoudre chacune de ses tâches: il faut un modèle distinct pour chacune d’elles. Dans ce paradigme, chaque tâche est traitée de façon indépendente, entrainant plusieurs conséquences, par exemple : le fait que chaque tâche soit traitée de façon indépendante ne permet pas le partage d’information entre elles. Nous allons travailler à la mise en commun de l’information tirée de différentes tâches afin de disposer de plus d’information pour le traitement des tâches futures.

L2M – U-Pro Soccer

U-Pro Soccer is a Canadian sportstech company developing a mobile-first training platform that uses AI and computer vision to help youth athletes practice soccer drills at home using only a smartphone and a training mat. Many young players lack access to structured training, especially outside of team environments, and U-Pro aims to fill this gap by offering real-time feedback on skill execution, agility, balance, and effort. This project will support the development of U-Pro’s scoring system by designing movement evaluation logic, integrating it into the AI feedback engine, and preparing validation plans to guide future testing. These efforts will strengthen the product’s accuracy and commercial readiness while ensuring that the feedback provided aligns with core principles of skill development. The expected benefit to the partner organization is a validated and defensible scoring framework that enhances the accuracy, usability, and market competitiveness of its AI-driven training product, helping accelerate its path to commercialization.

L2M Project – Salyx Medical Inc.

Salyx Medical is advancing a proactive health monitoring solution designed to protect seniors and individuals with chronic conditions who live alone, addressing the critical gap left by reactive tools like fall detectors or emergency buttons. The project focuses on overcoming challenges related to establishing strategic partnerships and executing real-world pilot studies—key steps in validating the technology and building trust in traditionally cautious healthcare environments. A major goal is to raise awareness within the senior care sector through targeted outreach, encouraging early adoption and easing the path to market. The project’s pilot run in a senior care facility will be instrumental in collecting user feedback, demonstrating clinical value, and refining the system for a successful market launch.

Technical and business intern(s) working within cross-functional teams to commercialize AI-powered solutions (3)

There is no overview on Umbrella applications so this section is left empty.

Développer une stratégie de gestion et d’évaluation des risques en biosécurité et biosûreté liée à la manipulation et au transport des agents pathogène de groupe de risque 2 au laboratoire NC2 d’EVAH, en conformité avec la réglementation en santé animale

EVAH Corp, est une entreprise de biotechnologies axée sur le développement des technologies dans le domaine de la santé animale, cherche à optimiser la gestion des risques liés aux agents pathogènes de niveau NC2 dans son laboratoire. Ce projet vise à développer une stratégie proactive d’évaluation et de contrôle des risques de biosécurité/biosûreté, conformément à la Norme canadienne sur la biosécurité. Les activités incluront l’analyse des procédures existantes, l’identification des lacunes, et la proposition de mesures correctives. Les retombées attendues incluent une meilleure conformité réglementaire, une réduction des risques pour les employés, le public et environnement, et une amélioration continue la gestion des risques et la pérennité d’EVAH Corp.

Empowering HR Decisions through Explainable Agentic Intelligence and Predictive Analytics

The current landscape of human resources faces significant challenges in streamlining job documentation and job evaluation processes, which are critical for consistent, efficient, and defensible compensation-related decisions. Laulima Consulting Inc., as the partner organization, aims to overcome this by developing an AI-powered tool that supports the end-to-end creation and evaluation of job descriptions. The core innovation challenge or improvement priority that Laulima needs to solve is to move beyond traditional, potentially manual, and inconsistent methods of job documentation and evaluation. This project proposes to address this by integrating market intelligence, conversational AI, and standardized frameworks to ensure greater consistency, efficiency, and defensibility in human capital decisions, specifically concerning job level and pay grade determination.

This project will help Laulima Consulting Inc. address these challenges in a way that extends significantly beyond typical day-to-day business operations by creating a standardized and scalable AI agent. This agent will not only synthesize internal inputs with external market intelligence to generate customized job descriptions but will also leverage a natural language AI agent for interactive validation and refinement of job content with managers. Critically, the project involves automating point factor job evaluation methodologies, where AI interprets job content against selected criteria, assigns point values, and provides justifications for transparency and auditability. This level of AI-based factor evaluation, combined with a flexible evaluation framework, dynamic visualizations, and consistency checks, significantly enhances HR’s strategic decision-making capabilities, moving beyond manual tasks to provide data-driven insights and a defensible system. Solving this complex problem requires specialized expertise in Natural Language Processing (NLP), Large Language Models (LLMs) and Agentic AI systems for content analysis and conversational AI, as well as data aggregation and benchmarking methodologies. Additionally, it necessitates advanced software engineering for building configurable and scalable systems, and deep domain knowledge in human resources, including job evaluation and compensation frameworks.

Étude des menaces au niveau des environnements des bornes de recharge pour les véhicules électriques

Alors que l’industrie des véhicules électriques connaît une croissance rapide, l’importance des stations de recharge ne peut être sous-estimée. Cependant, cette croissance s’accompagne également d’un problème urgent en matière de cybersécurité. Ce projet est dédié à la protection des communications entre les stations de recharge et leurs systèmes de gestion, en particulier ceux qui utilisent le protocole de communication OCPP. Les vulnérabilités potentielles de ce protocole pourraient être exploitées par des attaquants informatiques, entraînant des perturbations dans les réseaux de recharge et compromettant même la sécurité énergétique.

Ce projet adopte une approche innovante pour lutter contre ces menaces. Il vise à développer un système intelligent capable de détecter automatiquement les tentatives d’intrusion, en particulier celles qui ciblent les mécanismes d’authentification. Ce système utilisera des techniques d’intelligence artificielle de pointe, telles que l’apprentissage profond, pour analyser des données réelles et simulées. En identifiant les comportements suspects, ce système renforcera considérablement la sécurité de nos réseaux de recharge.

En rendant les réseaux de recharge plus sûrs, plus fiables et mieux protégés contre les cyberattaques, ce projet répond non seulement à un enjeu crucial pour l’industrie des véhicules électriques, mais contribue également à l’objectif plus large de transition vers une mobilité électrique durable.

L2M – MAP Medicine

Luma is an augmented reality (AR) surgical navigation platform designed to help surgeons see important blood vessels and anatomy directly on a patient during surgery. Right now, surgeons often have to rely on memory or markings to make decisions, which can lead to longer, riskier operations. Our project will explore who buys new surgical tools like Luma, what matters most to them, and how to introduce this technology successfully into hospitals and private clinics. By understanding buyer needs and decision-making processes, we’ll help MAP Medicine create a practical plan to launch Luma, making surgeries safer and faster while ensuring the company’s innovation reaches the patients and institutions that will benefit most.

Leveraging online violence prevention technologies to identify, understand, and counter violent extremism, gender-based violence, and mis/disinformation

Violent online content and radicalization have become so pervasive that it is contributing to the polarization of political discourse, a fraying of the social fabric, and to widespread human rights violations. Moonshot is devoted to ending online harms and protecting vulnerable individuals. It has developed and built technological tools to help governments identify and disrupt a wide range of violent threats, from terrorism to organized crime, child sexual exploitation, and trafficking, among other forms of abuse and violence.
Moonshot is seeking one Mitacs intern to support two of its ongoing research projects focused on countering radicalization, gender-based violence and misinformation, and assessing online threats. Activities include conducting research on online harms to assess when and how they are encountered by Canadians; who is seeking support or engaging with extremist content; and how violence prevention services can be made more accessible online, and engage vulnerable populations. The Mitacs intern will research and analyze relevant data, conduct desk research, and contribute to reports and campaigns. The projects will benefit the partner organization by supporting its ability to work with governments and local behavioural health services to create skilled provider networks, design effective online outreach strategies, and improve service delivery in virtual settings.

L2M – AI Driven Predictions for Safer Drug Use in Pregnancy

FetoSafe is an AI-powered startup designed to predict whether small-molecule drugs can cross the placental barrier and reach the fetus. This remains a major challenge in maternal healthcare, as most medications lack reliable data on fetal exposure.
FetoSafe uses only the SMILES representation of a compound, capturing its chemical structure, to estimate the risk of placental drug transfer. Trained on a curated set of known compounds, the model provides a fast, scalable, and non-invasive way to support safer, more informed decisions in drug development during pregnancy.

Enhancement of a Pharmacogenomic Treatment Decision Support Tool for Schizophrenia – Year two

The GeneSight (GS) Psychotropic Test (AssureRx Health, Inc.) measures individual variations in drug-related genes and recommends optimal drug choices for individual patients. A clinical study will be conducted to validate novel gene markers of antipsychotic-induced weight gain for use in an enhanced version of GS. Sarah Groleau will be responsible for genotyping DNA from 600 schizophrenia patients to evaluate which of the significant gene variants (UofT/CAMH) predict the most troubling side effect of antipsychotic medications: weight gain and metabolic syndrome. She will also be involved in economic analyses of financial burden in failed treatments, with MaRS EXCITE, to position the GS tests for OHIP reimbursement. In addition, Sarah will shadow ARxC personnel, gaining exposure to all aspects of the company’s function. Sarah’s project will provide AssureRx Canada with substantial intellectual capital and access to research expertise at CAMH, enabling accelerated timelines for AssureRx product release and Ministry of Health licensing.