Innovative Projects Realized

Not All Backports Are Equal: Analyzing the Risk/Value Trade-offs in Backport Acceptance Decisions

Ensuring the long-term stability and security of widely-used software is a critical operational challenge. A key part of this is “backporting,” the high-stakes process of deciding which updates to apply to stable product versions. Currently, developers often make these crucial risk-versus-value judgments without formal, data-driven guidance. This project will develop an intelligent decision-support framework to solve this problem. By applying state-of-the-art machine learning to historical project data, our system will learn to automatically classify backport proposals, profile their risk and value, and provide actionable recommendations. For the participating institutions, this project pioneers new techniques in AI-driven software engineering, culminating in a tangible decision-support tool. The ultimate benefit is a direct contribution to enhancing software reliability and developer productivity, strengthening the innovation capacity of the wider technology sector.

Exploring the social learning potential of community-based participatory sustainable fashion and textiles activities framed by speculation.

This internship supports the creation of an open-access publication on sustainable fashion, intergenerational learning, and craft practices. It investigates how participatory speculative design through Fashion Fictions workshops enables community textile groups to collaboratively imagine alternative fashion futures, fostering collective imagination, social learning, and agency. The project is the first Mitacs Globalink partnership between Nottingham Trent University’s Sustainable Transitions Research Group (STRG) and Concordia University’s Geography, Planning, and Environment community, bridging sustainable fashion/design and decolonial geography. The research environment provides hands-on experience in workshops, intergenerational sessions, and co-design exercises, with access to STRG networks, seminars, and interdisciplinary collaborations. The internship will support the intern’s development, contribute to an open-access publication, strengthen cross-institutional collaboration, and advance Canada’s leadership in socially grounded sustainability research.

Analytical Assessment of Bond Resistance and Failure Modes of Post-Installed Bundled GFRP bars in Concrete

Glass Fiber Reinforced Polymer (GFRP) bars have become increasingly important in civil engineering due to their exceptional resistance to corrosion, high strength-to-weight ratio, and durability when compared to traditional steel reinforcement. This has led to widespread adoption, particularly for the rehabilitation and strengthening of existing structures. Previous research has explored the mechanical performance and bond behavior of single, cast-in, and post-installed GFRP bars, providing valuable insights into how factors such as embedment length, bar diameter, and adhesive formulation affect anchorage capacity and structural integrity.
Much of the literature, however, remains focused on single-bar scenarios or post-installed steel bars; studies involving bundled post-installed GFRP bars are rare, despite these systems being increasingly applied in real retrofit projects where space or design requirements necessitate bar bundling.
The main goal of this research is to improve the analytical understanding of bond resistance and failure modes in post-installed bundled GFRP (Glass Fiber Reinforced Polymer) bars for concrete retrofit applications. The work is the central focus of my B.Eng. thesis and aims to connect experimental findings with finite element (FE) simulations to help develop practical recommendations for future design and safer construction practices.

Neural Oscillations During Sleep: Mathematical Modeling and Data Analysis

When we sleep, the brain produces rhythmic patterns of activity that help us learn, remember, and maintain healthy function. Some of these patterns travel across the brain like waves. Scientists believe these traveling waves are important for how the brain processes information, but we still do not fully understand how they form or what they do.

This project brings together researchers in Brazil and Canada to study these neural oscillations. Our team will use computer models and mathematical techniques to simulate how brain areas interact to generate wave-like patterns. We will also analyze empirical recordings of brain activity to detect when and where these oscillations occur. By combining these two approaches, the project aims to uncover the basic principles that shape neural oscillations during sleep.

Understanding these patterns is significant because disruptions in sleep rhythms are linked to conditions such as Alzheimer’s disease, epilepsy, and mood disorders. Insights from this research may help guide future studies on brain health and inspire new ways to understand — or potentially correct — abnormal brain activity.

Analyzing Explanatory Texts from the Perspective of Functional Discourse Grammar

This project focuses on how people use modifiers to explain abstract terms in English and Chinese, employing Functional Discourse Grammar (FDG) to describe the linear positions of modifiers, the levels of meaning they target, and whether they affect truth conditions. Aimed at education, public communication, and language technology, it will deliver clear hierarchy–scope–linearization criteria and aligned annotation guidelines, with potential benefits across these areas. It has especially strong potential in communication-intensive domains such as law, healthcare, and narrative services. The project will use multilingual corpora and learner data to drive theoretical development, potentially advancing Canada’s multilingual ecosystem.

Novel techniques to refactor software systems using reinforcement learning

Refactoring is an important software development activity that employs various techniques to enhance the structure and quality of source code without altering its functionality. Over the last two decades, researchers in the field have proposed several tools and techniques to enable automated refactoring. Despite a plethora of literature, though some steps can be automated, overall refactoring remains a manual process. In this project, we aim to develop a method to generate refactored code from any given code to improve its maintainability using reinforcement learning.

Machine Learning-Based PV and Load Forecasting for Power System Optimization: Using real data in Cambodia with IEEE 9-Bus Validation

The growing number of renewable energy sources in Cambodia, such as solar farms, makes it more difficult to schedule power system operations to satisfy demand and save operating costs. Forecasting is one way to manage energy. An excessively high prediction accuracy indicates that managing the spatial energy in Economic Dispatch (ED) and Optimal Power Flow (OPF) is simple. One graduate intern will travel to Concordia University to conduct advanced research on machine-learning-based photovoltaic (PV) and load forecasting models using data from Electricité Du Cambodge (EDC) in Cambodia and the weather variable chosen from public data from NASA POWER in collaboration with the IEEE 9-Bus. The expected outcomes are improved forecasting accuracy, reduced grid power losses, reduced energy costs, and increased use of renewable energy. The findings will support Cambodia’s transition to clean energy and aid in the planning of a more reliable and sustainable power system. The project will also help Canada become a leader in renewable energy innovation, improve cooperation between Cambodia and Canada, and exchange new research findings.

Use of bio-inputs in the ecological restoration of iron ore mine sites in northern Quebec, Canada

This research project focuses on restoring the land that has been disturbed by post-iron-ore extraction sites at the Timmins 7 site in Newfoundland and Labrador Province and the other at the Good Wood site in Quebec Province at Tata Steel Minerals Canada Ltd. (TSMC) in Schefferville. These post-mining sites have extreme climatic conditions of a harsh subarctic climate, like long, harsh winters, low temperatures, nutrient-poor soils, and short growing seasons, which create substantial challenges for vegetation establishment and make it difficult for plants to grow naturally. To solve this problem, the project will test the use of “bio-inputs,” such as beneficial microbes—fungi and bacteria that naturally help seedling roots absorb nutrients, tolerate stress, and survive in harsh environments. We will grow native boreal seedling species like Picea mariana, Betula glandulosa, and Alnus viridis ssp. Crispa in the greenhouse with and without these microbes and then plant them in two active mine restoration sites (Timmins 7 and Good Wood). By monitoring plant survival, growth, soil quality, and nutrient recovery, we aim to identify the best combination of soil amendments and microbial treatments that can speed up ecological restoration of the post-iron-ore-mining area.

Investigation of strength and performance evaluation of hybrid wide flange (WF) beam-to-box column connections with inner diaphragms

In modern structural engineering, the connection between beams and columns plays a critical role in determining the overall performance and safety of steel structures and using an inner diaphragm is to improve strength and load transfer from beam to column too. This connection of WF beam to box column with inner diaphragm is responsible for transferring loads such as bending moments, shear forces, and axial forces between the members. The study of hybrid wide flange beam to box column connections with inner diaphragms has attracted increasing attention because of their superior load-carrying capacity, ductility, and ease of fabrication in high-rise and seismic-resistant structures. We usually use this ideal for high-rise buildings, bridges, and industrial facilities to create strong, load-bearing frames and seismic-resistant structures as well. In this study, the analytical assessment of mechanical behavior such as failure modes of the connection, evaluate the strength, load-displacement and the parametric effect of diaphragm properties by using numerical simulation software called ABAQUS/CAD. The finite element analysis results showed that variations in diaphragm parameters had only a minor effect on the overall load capacity and strength. It has showed us that parametric study is very important to improve strength and failure mode of connection.

AI Agent–Driven Digital Twin Framework for Intelligent Simulation and Optimization of Demand Responsive Transit (DRT) Systems

This project aims to develop an optimal Agentic AI Framework for next-generation intelligent mobility systems. Rather than focusing on a single algorithm or application, the research seeks to establish a generalizable architecture that integrates advanced optimization methods with graph-based large language model orchestration. Demand-Responsive Transit (DRT) serves as the primary testbed due to its inherent operational complexity, but the framework is designed to support a broad range of urban mobility services.

The core contribution lies in unifying three methodological components:
(1) an agentic control layer that interprets natural-language inputs and coordinates multi-step reasoning using a LangGraph-based structure;
(2) an optimization layer that adapts routing, dispatching, and policy parameters through formal algorithmic models; and
(3) a digital-twin simulation layer that evaluates system behaviors within an interactive urban mobility environment.

By integrating linguistic reasoning with algorithmic optimization, the project addresses a key gap between human decision intent and computational execution in current mobility systems. The resulting framework offers a scalable and extensible foundation for analyzing complex transportation scenarios, supporting evidence-based planning, and advancing research on AI-driven urban mobility.

TruEffect Digital Data Workflow, Exchange, and Analysis Innovation Project

This project aims to modernize Trueffect’s end-to-end data workflow by designing and enhancing a secure digital platform for client data exchange, validation, and analysis. The organization currently handles complex analytical datasets from multiple client partners, requiring a more efficient and scalable approach to ingestion, quality assurance, and internal analyst workflows. The proposed initiative introduces a structured and innovative system that enhances operational efficiency, strengthens data governance, and supports future analytical capabilities.

The intern will work with the academic supervisor and Trueffect to advance a secure web-based platform that integrates authentication, file ingestion, exploratory data analysis (EDA), and structured data-quality reporting. The project includes mapping current processes, identifying inefficiencies, and implementing improvements that streamline client uploads, reduce manual analyst workload, and increase reliability. The intern will contribute to architecture refinement, feature implementation, and workflow optimization, including enhancements to validation routines, shared-project synchronization, and the end-user experience.

IoT device identification using continuous machine learning models

In this project, we will design and develop an IoT device auto-detection model and deploy it on a Raspberry Pi. The model will follow a continuous machine-learning approach and will be trained using an IoT dataset developed in our lab. The Raspberry Pi will function as a Wi-Fi router, allowing IoT devices to connect to it. Upon connection, the system will automatically identify each device—including its vendor and model—in real time.