Projets novateurs réalisés

Building a semi-supervised machine learning model to predict biomolecular condensates

Princess Margaret Cancer Centre belongs to the University Health Network (UHN), Canada’s leading biomedical research organization. The Centre focuses on cancer research across various fields, including genomics, informatics, signaling, health services, and biophysics.
Dr. Kumar’s lab is currently investigating the consequences of genomic alterations in intrinsically disordered regions (IDR). IDRs are present in proteins that undergo liquid-liquid phase separation (LLPS) and form biomolecular condensates [1]. IDRs lack a fixed structure yet play vital roles in cellular function [2]-[4]. Genetic alterations can disrupt biomolecular condensate activity, leading to neurodevelopmental disorders and cancer [5]. Despite their biological significance, IDRs are often overlooked in drug discovery. Current experimental methods to identify IDRs lack throughput. Therefore, this project aims to address these challenges by developing an advanced in-silico approach to predict LLPS proteins.
As a research assistant, the intern will take on this project and contribute to cutting-edge computational research. If successful, this in-silico approach could significantly reduce reliance on costly experimental procedures while accelerating breakthroughs in precision medicine. By enhancing predictive efficiency, this computational technique could open new avenues for drug discovery, enabling the targeted intervention in disordered proteins. Furthermore, identifying key LLPS proteins could provide deeper insights into cancer biology.

Scaling Tabular-Timeseries Foundation Models for Large-Scale Financial Data

TD Bank, as a leader in financial services, relies on predictive modeling to improve customer insights, risk management, and fraud detection. However, current machine learning approaches struggle to scale effectively across TD’s vast transactional datasets, leading to challenges in handling heterogeneous financial products, long-term forecasting, and multi-task learning. This project aims to address these challenges by developing scalable tabular-timeseries transformer architectures capable of learning from hundreds of millions of transactions across diverse financial services. By integrating self-supervised learning and multi-task optimization strategies, TD Bank will benefit from:
(1) Improved predictive performance in account acquisition, customer retention, fraud detection, and delinquency
prediction.
(2) Reduced model fragmentation, allowing a single scalable model to handle multiple financial objectives.
(3) Operational efficiency, decreasing computational costs by consolidating several models into a unified framework.
(4) Improved explainability, ensuring compliance with financial regulations and risk assessments.
This research aligns with TD’s commitment to AI innovation and financial technology advancements, providing direct business value through AI-driven decision-making and enhanced risk management.

Venn 2025 Summer Intern

At Venn, we’re not just building a product—we’re revolutionizing how Canadian businesses operate. As a rapidly growing fintech startup serving over 3,000 Canadian businesses, our ambition is to streamline our internal processes as we scale towards supporting over 10,000 companies. To achieve this, we’re looking for a motivated individual to take ownership of high-impact projects that eliminate manual processes and drive measurable operational efficiency.

Scaling USA-NPN Sampling Design using LLM Agents

This project, part of the Global AI Alliance for Climate Action, aims to improve how the USA National Phenology Network (USA-NPN) collects and balances seasonal plant and animal data. By using large language models (LLMs) and AI-driven workflows, the project will help guide citizen scientists toward underrepresented species and locations, addressing gaps in the current dataset. This will make USA-NPN’s data more complete and useful for tracking climate change impacts. For Vector Institute, this collaboration showcases AI’s role in solving real-world environmental challenges, reinforcing its leadership in responsible AI innovation.

Real-Estate Industry Transformation Through Innovative AI Platform Development

Royal Trillium Homes Inc. is a startup focused on transforming the real estate industry through creating an innovative platform and data-driven solutions. Through extensive market research, incorporating insights from PricewaterhouseCoopers (PwC), Deloitte, and direct feedback from clients, agents, and brokerages, critical inefficiencies exist. Currently the real estate industry relies on outdated processes, leading to mismatched client-agent pairings, limited agent visibility, and inefficient brokerage recruitment. Clients depend on personal referrals and static listings, while agents struggled to stand out in a crowded market. Brokerages faced hiring challenges due to traditional recruitment methods lacking real-time performance insights. This project will bring Information Technology expertise that currently does exist within Royal Trillium Homes, to develop an Artificial Intelligence (AI)-powered platform designed to eliminate inefficiencies in client-agent matchmaking, enhance agent visibility and streamline brokerage recruitment. Addressing these inefficiencies requires expertise of an intern with knowledge in AI, data science, user experience design, and real estate market dynamics, along with collaboration across industry and technology sectors. The fragmented nature of real estate transactions and the slow adoption of digital solutions further compounded these challenges, highlighting the need for a more transparent, efficient, and scalable industry framework.

Integrating AI with ITSM for Smarter Support and Workflow Automation

This project aims to develop and apply AI technology within an IT Service Management (ITSM) system to improve IT support and efficiency. By training Freshservice’s Freddy AI using real data, we will enhance the accuracy of ticket handling, streamline repetitive tasks, and automate routine IT workflows. This will significantly reduce the workload of IT staff, allowing them to focus on important strategic tasks. Co-op students involved in the project will gain hands-on experience in AI implementation and automation, directly linking their academic knowledge to practical industry challenges. Ultimately, this initiative will result in faster IT response times, better user satisfaction, and improved efficiency for the partner organization.

L2M_Marketing Strategy for VT-Patch, a medical smart patch as RPM system.

The VT-PATCH project addresses a crucial gap in pediatric healthcare by introducing a user-friendly medical patch
designed for short-term, continuous monitoring of infants and young children. Unlike the current complex system of
multiple devices and wires, the VT-PATCH offers an all-in-one solution to measure vital signs like heart rate,
respiratory rate, blood oxygen saturation, and body temperature. This innovative approach not only improves the
quality of care for sick children but also enhances operational efficiency in healthcare facilities. By facilitating earlier
patient release, the project increases bed availability, reduces nursing requirements, saves costs, minimizes energy
consumption, and optimizes time management. The VT-PATCH stands out in the market as a tailored solution for
pediatric patients, ensuring effective remote monitoring from an early age and contributing to the advancement of
pediatric healthcare in Canada.

European Responses to China’s Support for Russia in its War against Ukraine: A Study of the Czech Republic, Poland, and the Baltic States

This research project is situated within the broader scientific context of international relations, security studies, and European foreign policy. In recent years, the geopolitical landscape has been significantly shaped by China’s increasing global influence and its complex relationships with both Russia and the West, and it changed even more drastically when the Russia has started its full-scale invasion to Ukraine on February 24, 2022. The study would fill a gap in the literature on Europe-China relations in the context of the ongoing war in Ukraine. While most existing analyses focus on the EU’s collective stance or major Western European powers like Germany and France, this project highlights the often-overlooked perspectives of the Central-Eastern Europe (CEE) countries that are among the most vocal critics of Russian aggression. Relatively little or no scholarship has focused on how specific CEE countries, particularly the Czech Republic, Poland, and the Baltic states, perceive and react to China’s indirect support for Russia. By examining potential variations in the countries’ responses—particularly in light of their differing levels of support for Ukraine—the research aims to identify patterns and explanations behind these differences.

An investigation into the failure trends of an AI tool for the detection of tumour margins

Breast cancer is a global health challenge, affecting millions of women annually. In 2020, it was the most common cancer in 109 countries, including Canada. Breast-conserving surgery (BCS), or lumpectomy, is the standard of care for early-stage breast cancer. The goal of BCS is to remove malignant tissue while preserving healthy tissue but achieving tumor-free margins remains a significant challenge. Permanent histopathology, the gold standard for margin assessment, typically takes 2–5 days, leading to reoperations in over 20% of cases due to positive margins.

Perimeter Medical Imaging AI, a Toronto-based medical device company founded in 2013, has developed an OCT device with an embedded AI decision support module. This tool provides real-time feedback on tumor margins during surgery, potentially reducing reoperation rates. However, the AI module’s performance is not uniform across all patient subgroups and device conditions. A retrospective efficacy study of the AI tool has been published in a peer-reviewed journal, but further investigation is needed to understand its limitations and improve its reliability.

This project will focus on identifying trends in the AI tool’s failure modes, including false positives and false negatives, across different demographics, disease types, and device variability. By addressing these limitations with changes to the annotation or training data composition, we aim to enhance the tool’s accuracy and usability, ultimately improving patient outcomes.

Réingénierie d’un processus de classification de la qualité de placages de bois

Ce projet de recherche consiste à classifier automatiquement du bois plaqué selon 16 catégories à l’aide d’apprentissage automatique. Il sera réalisé en collaboration avec une entreprise, R.Perron, qui collaborera activement pour ce projet en fournissant des données réelles de production ainsi que l’avis d’expert dans ce domaine. Actuellement, cette classification est faite visuellement par un opérateur. Étant donné le grand nombre de catégories (16), l’entreprise souhaite automatiser ce processus, car celui-ci est laborieux et difficile pour l’opérateur. De plus, des erreurs de classification peuvent causer un retour des produits de l’entreprise par ses clients, ce qui emmène des pertes. Aucune donnée n’est actuellement disponible, donc l’objectif de ce projet sera à la fois d’entraîner des modèles de vision pour effectuer cette classification ainsi que d’effectuer la collecte des données, le tout dans un contexte réel de production. Ce projet a donc une nature exploratoire, car nous serons emmenés à tester différentes caméras ainsi que différents positionnements de celle-ci.

Hazelnut alley cropping for the Pacific Northwest

Commercial hazelnuts are a woody perennial crop that requires significant upfront investment. They are typically grown in a monocropped orchard, with bare soil or a closely-cropped grassy orchard floor. Hazelnuts are usually harvested from the ground, meaning that the presence of intercrops is a hindrance for harvest. This results in lower in-field diversity, wasted production space, and unrealized income potential. The use of alley space for production has the potential to increase food and nutrient yields as well as economic returns from the system, but management strategies for doing so without interfering with nut harvest have yet to be established for the Pacific Northwest.

We will establish a hazelnut alley-crop system with the goal of developing strategies for allowing both nut and intercrop harvest. We plan to evaluate the feasibility of intercropping with hazelnuts under three different canopies – single stem, multi-stem, and hedging forms. Strategies may include using harvest technologies such as straddle-harvesters and ATV-pulled harvesters that work in narrow spaces. We will also experiment with different intercrops to develop best practices with the goals of with the goals of allowing efficient nut harvest, maximizing food production from the intercrop, and maintaining soil health and fertility under organic management.

Development of new solid lubricant coatings for different industrial applications

This project focuses on the development, testing, and validation of next-generation solid lubricant coatings for cutting tools designed for hard-to-cut materials commonly used in aerospace and automotive industries. Traditional machining processes often rely on oil- and water-based lubricants, leading to excessive waste, environmental contamination, and increased operational costs. By introducing advanced solid lubricant coatings, this project aims to enhance machining efficiency while reducing industrial waste and water consumption.
The research is centered on improving tool performance in demanding machining applications. The coatings are engineered to minimize friction, reduce heat generation, and extend tool life, ultimately optimizing productivity. Hard-to-cut materials, such as heat-resistant alloys and composites, pose significant challenges in conventional machining. These coatings provide a dry, eco-friendly solution that improves cutting efficiency and also decreases reliance on liquid lubricants.
By addressing critical challenges in machining technology, this initiative supports a transition toward greener, more sustainable manufacturing practices. The outcomes have the potential to benefit industries seeking high-performance solutions that align with environmental sustainability goals while maintaining production efficiency.