Innovative Projects Realized

L2M – Development of a user-driven, evidence-informed, AI-powered post-pregnancy support application

The first year after pregnancy is a critical but often overlooked by healthcare systems globally. Many people face physical, emotional, and social changes, yet there is very little formal support. In Canada, most people receive only one six-week follow-up appointment, and support varies depending on who their care provider was. As a result, people with lived experiences of pregnancy turn to online sources, family, or friends for advice. This fragmentation in care leads to missed warning signs, preventable complications, and increased strain on emergency services. To help address this gap, this project will support the development of Hey Aunty!, a novel digital tool that uses artificial intelligence (AI) to provide personalized, culturally sensitive, evidence-based support to people during the first year after pregnancy. It is designed to work alongside medical professionals and not replace them. As a first step, this internship will focus on designing and testing the feasibility of the app’s core feature: an AI-powered conversational companion for tech-savvy, English-speaking individuals who have experienced a healthy, full-term birth. To achieve this, the intern will conduct interviews with end-users and experts. These will also help curate and train a domain-specific AI model that underpins the conversational companion.

L2M – AllerEase: Personalized Allergy-Safe Grocery Tool

AllerEase is a personalized grocery recommendation tool that helps allergy-sensitive users shop safer and easier. It generates user-specific shopping lists based on allergy profiles and live product data. This project aims to refine matching logic and validate usability through real-world testing.

L2M – Characterizing the pathology of airway stenosis in canine patients: a first step to advancing airway stenting.

This project involves histopathological analysis of canine airways with stenosis, a type of obstructive disease that impedes airflow. This research is aimed to characterize the disease and understand the cellular mechanisms that drive the disease progression, and therefore inform the future directions of treatment innovation.

L2M – Load Forecasting Improvement in Smart Grids

This project focuses on improving short-term electricity demand forecasting, which is essential for ensuring the reliable and cost-effective operation of the power grid. Many Canadian utilities are now using demand-side management strategies such as peak shaving, reducing electricity consumption during peak hours, to lower costs and reduce strain on the grid. However, these strategies can unintentionally distort electricity consumption data, making it more difficult to accurately forecast future demand using traditional models.

The project proposes a novel, software-based forecasting solution that detects and adjusts for the effects of peak shaving to address this challenge. The approach uses advanced machine learning techniques to incorporate key indicators, such as the timing and duration of peak shaving events, into the forecasting process. This results in more accurate and robust predictions of electricity demand, even when smart grid interventions have altered the data.

The improved forecasting model will help utilities like Saint John Energy plan more efficiently, reduce their reliance on fossil-fueled backup systems, and better integrate renewable energy into the grid. These outcomes support Canada’s broader goals for environmental sustainability, grid modernization, and energy affordability.

The project contributes to building smarter and more resilient electricity systems across Canada in the long term by equipping utilities with advanced tools to manage clean energy transitions effectively.

Development of rapid and accurate genomic techniques for ballast water UV treatment

The United States Coast Guard (USCG) recently introduced stringent regulations for the treatment of ballast water. Ultra-violet (UV) light is a useful technology in a ballast water treatment system (BWTS), for inactivating species which could be invasive and harmful to humans and the environment. UV damages DNA and prevents replication, but the vital stain methods mandated in the USCG protocol do not detect UV damage. Alternative culture-based measures of cellular replication capacity are yet to be approved, time consuming, and have limitations (some species mayn’t grow). Rapid and accurate assays to identify cells that are UV damaged beyond recovery are critical for the acceptance of UVbased BWTSs. Hence practical methods based on genomics/transcriptomics will be developed for rapid, high throughput, on-site assessments of DNA damage related to key functional genes in micro-eukaryotes. TROJAN Technologies will use this technology to validate UV sterilization for ballast water treatment at the global scale.

L2M – Next-Generation Multisource Thermo-Photovoltaic Receivers for Wireless Power Transmission Networks

We’re developing the first high-efficiency, ultra-long-range, multi-source thermophotovoltaic (TPV)-based wireless power transmission (WPT) receiver, designed to convert laser beams into electricity without relying on fragile PV surfaces. It uses a spectrally selective absorber to convert laser into tailored thermal emission, matched to low-bandgap GaSb cells. Unlike traditional laser WPT systems, it performs effectively in space, underwater, harsh environments, and ultra-long distances, where cables, batteries, or PV-based WPT fail or lead to high losses, costs, and oversized infrastructure. With low beam divergence, low atmospheric attenuation, spectral and laser flexibility, uniform high-power density tolerance, thermal buffering, and photon recycling, our system de-livers unmatched efficiency, scalability, and mission adaptability. From drones and satellites to deep-space and subsea infrastructure, this technology redefines untethered power delivery across Earth, orbit, and beyond.

L2M – LeanPrompt: Accelerating Generative AI with Smarter Resource Utilization

We aim to develop and implement a cost-efficient framework for communicating with proprietary generative AI platforms such as ChatGPT. In fact, these provider companies expose their models through an interface that can be accessed via API. However, calling their API will incur a cost considerably based on our request. Basically, smaller models are cheaper than larger models. However, the smaller models are less capable and may not generate accurate responses. Hence, we aim to reduce the cost while maintaining the accuracy and latency of our product. We will implement this approach with three mechanisms. First, we compress input as the larger queries will charge us more (It is calculated per number of words). Secondly, we use a routing mechanism to use smaller models for simpler tasks and larger models for more complex tasks. Lastly, we will use a caching mechanism to leverage the previously answered data and avoid invoking the models every time. All these approaches would reduce our cost, and also they can preserve the performance of our model with smart techniques. At the end, the partner organization can invest its budget more broadly across different objectives. Also, they can engage more user on their platforms by providing accurate responses.

L2M – NeurolAssist

We are developing a specialized AI-driven Clinical Decision Support System (CDSS) for neurology that integrates a large language model with domain-specific tools (such as neuroimaging analysis, EEG interpretation, and up-to-date medical knowledge retrieval). The system serves as an intelligent assistant for neurologists, helping analyze patient data (e.g. MRI scans, EEGs, clinical notes, etc.) and providing evidence-based diagnostic support. Uniquely, our LLM-powered solution engages in dialogue with the doctor and asks clarifying questions when it’s not sure or needs more information, and it continuously checks the reliability of its tool. We aim to integrate the system into electronic medical records (EMRs) used in Canada and the US to potentially reduce diagnostic errors and save physicians time in neurological care.

L2M – Commercializing CORIMA: A Systems-Based Platform for Interorganizational Risk Management in Ports

This project will support the early development of CORIMA, a new digital platform designed to help ports and other industries improve how different organizations work together to manage safety and risk. The platform is based on research done at Dalhousie University and uses a systems-thinking approach to identify gaps in coordination that can lead to delays, accidents, or inefficiencies. Through interviews, early testing, and business planning, this project will help turn academic knowledge into a practical tool. The expected benefit to CORIMA is a stronger foundation to move forward with commercialization and expand its use across the port sector and other industries.

L2M – A Novel Robotic Tool for Material Removal Applications Using Magneto-Rheological Actuators

**Public Project Overview**
This project will develop a new type of robotic tool for sanding, grinding, and polishing using advanced magneto-rheological (MR) actuators. Unlike traditional tools that rely on air pressure or electric servo motors, the MR-based tool can quickly and smoothly adjust how hard it pushes on a surface, making it ideal for high-precision jobs. The goal is to turn an existing lab-tested prototype into a production-ready product that can be used in real industrial settings. This tool will help our partner organization offer a next-generation solution to manufacturers looking to improve product quality, reduce manual labour, and automate surface finishing tasks more efficiently.

L2M Launch / Qc Fall 2025 / mTatt

mTatt is a medtech startup developing functional microtattoos for real-time, at-home health monitoring. Using dissolving microneedles — biocompatible structures under 1 mm in length — we painlessly deliver fluorescent sensors into the skin, creating a functional tattoo. These sensors respond to key biomarkers like NT-proBNP, glucose, or electrolytes, providing clinically valuable insights without blood draws or clinical visits. Data is captured via a smartwatch-style wearable and sent to the user’s phone, enabling patients and physicians to track health conditions continuously and intervene early. Our platform makes the diagnostic power of a lab accessible in real time, from the comfort of home.

L2M – DeepSimu: Next Generation of Deep Learning Frameworks

DeepSimu, the next generation of Deep Learning (DL) frameworks, is proposed to deal with the challenges which Artificial Intelligence (AI) practitioners are facing nowadays when using the available Machine Learning (ML) tools such as Tensorflow, Pytorch, Keras and ScikitLearn. While these frameworks necessitate high level python programming skills like object-oriented programming, DeepSimu has made a huge shortcut in the process of learning AI by eliminating Python from it. This novel framework presents a block diagram environment where different elements of a deep learning model such as Dense layers, Convolutional layers, Recurrent layers, and Attention layers are represented in the form of some blocks. The user can easily arrange these blocks according to the desired architecture and link them by drawing some lines between them. While Canada is considered as the birthplace of DL and modern AI in the world, it plays no significant role in the business of DL frameworks. As countries around the world race to leverage artificial intelligence (AI), a recent study found that Canada is falling behind. In a recent study from KPMG, Canada ranked 44th in AI training and literacy out of 47 countries, and 28th among 30 advanced economies. The study surveyed over 48,000 people in 30 advanced economies and 17 emerging economies, including 1,025 people in Canada. DeepSimu is believed to be a milestone for Canada in the business of AI. It is expected to facilitates and accelerates the process of learning and teaching AI in Canada. This innovation will result in increasing the number of people knowing how to work with AI in Canada and generally speaking would make a dramatic change in the public confidence toward AI.