Innovative Projects Realized

Assessing Railway Maintenance Intervals and Slow Order Management in a Low Traffic Climate-Sensitive Freight Railway Corridor

The impacts of permafrost thaw on northern railways can lead to temporary speed restrictions (“slow orders”) imposed to mitigate operational risk until permanent remediation works are completed. These slow orders cause unpredictability in nominal train operating speeds and ultimately affect the reliability of long/heavy trains operating across successive slow-ordered segments. Little is known about how slow orders should be managed on low-traffic railways subject to events such as permafrost thaw, which cause track degradation independent of accumulated tonnage or elapsed time.
This work will leverage track geometry data and track maintenance data to develop a data-driven framework to help low-traffic freight railways manage slow order occurrences and improve network reliability. The results will advance railway carriers’ understanding of effective slow order management under evolving climate conditions and will provide railways with transferable tools to forecast unplanned maintenance.

Fernwood Agentic Knowledge Graph for Housing Innovation

Canada’s housing crisis has prompted both government and industry to call for technological innovation in the building permit and development approval process, where regulatory complexity, fragmented policy documents, and inconsistent interpretation contribute to costly delays. Fernwood Agentic is responding to this need by building an AI system that helps architects, planners, and developers navigate municipal housing and zoning policy more quickly and accurately. This Mitacs project focuses on a core component of that system: a knowledge graph that captures the rules, zones, restrictions, and cross-references found across the City of Victoria’s policy documents and represents them as a structured, queryable network. Two interns will catalog Victoria’s policy corpus, design a schema defining the types of entities and relationships it contains, build an LLM-based pipeline to extract that information from the source documents, populate a Neo4j graph database, and validate the results with domain experts including architects, planners, and City of Victoria staff. The expected benefit to Fernwood Agentic is a set of core technical assets, including the populated knowledge graph, a reusable extraction pipeline, and a formal schema specification, that will improve the accuracy and reliability of its RAG system and serve as the blueprint for expanding to additional municipalities across British Columbia.

Trauma Focused Therapy Program: A Retrospective Chart Review

This project will help the Trauma-Focused Therapy Program (TFTP) at the Jewish General Hospital better understand who is accessing its specialized services for complex trauma and how patients’ symptoms change over the course of treatment. Complex trauma, which is often linked to repeated childhood maltreatment, can lead to serious mental health, physical health, and psychosocial challenges, yet specialized public services remain limited in Canada. Since its foundation in 2019, the TFTP has collected valuable clinical information through validated questionnaires and written feedback mechanisms that have not yet been systematically organized into a research database, allowing statistical modelling. Through this internship, a database will be created from patient files, and preliminary analyses will examine patient profiles and treatment outcomes. The results will help the program improve services, identify gaps in care, and strengthen trauma-informed practices. Ultimately, this work will support the recovery of Quebec complex trauma patients and inform evidence-based service provision in Canada and beyond.

Building a Location-Aware Mobile Application to Connect Consumers with Local Businesses and Community Incentives

This project will develop and pilot-test a mobile application that helps consumers discover local businesses, promotions, and community events in their neighbourhood. Built for Obodo Development Corp., the app will give small businesses an affordable alternative to large digital platforms that charge high commissions, offering simple tools to attract and engage nearby customers. Over eight months, two interns will take Obodo’s existing designs and technical plans and turn them into a working product, building the mobile interface, backend systems, location-based discovery features, and incentive redemption tools, then test the app with real businesses and consumers in a structured community pilot. The project will provide Obodo with a validated, pilot-ready product and a strategic roadmap for public launch, positioning the company to bring an accessible digital engagement solution to local businesses across Vancouver and beyond.

Medzy – Scientifiques en Action – Dossier patient intelligent en pharmacie – Couche IA vocale

Dossier patient intelligent en pharmacie : intégration d’une solution d’IA vocale et interconnectivité avec modules IA existants

Medzy Inc. développe des solutions numériques visant à améliorer l’accès, la qualité et la continuité des soins en pharmacie communautaire. L’entreprise conçoit des outils facilitant la gestion de l’information clinique et la communication entre professionnels de la santé. Ses solutions s’inscrivent dans une approche centrée sur le patient. Medzy collabore avec des acteurs du secteur de la santé.

Le stage porte sur le développement d’un dossier patient intelligent destiné au contexte pharmaceutique. Le ou la stagiaire travaillera sur l’intégration et l’analyse de données cliniques hétérogènes. Le projet inclut l’amélioration de modèles d’aide à la décision et l’évaluation de leur pertinence clinique. L’objectif est de renforcer la cohérence, la fiabilité et l’utilité des données patient.

Research Agent Evaluation Harness and Methodology

Tiptree builds AI research assistants for ML scientists. These agents help users discover relevant prior work, validate claims, assemble evidence with citations, and produce high-trust research artifacts (e.g., literature reviews, benchmark comparisons, technical memos).
A core bottleneck in deploying research agents is evaluation: unlike standard software, agent behavior is nondeterministic, multi-step, and tool-dependent, and failures often look superficially plausible (e.g., missing critical papers, hallucinated citations, or silent omissions). This makes it hard for both researchers and product teams to trust outputs and to iterate quickly on system design.
This project will develop a rigorous evaluation harness and methodology for research agents, grounded in best practices from agent evaluation (e.g., task/trial/graded-outcome framing, transcript analysis, pass@k and pass^k reliability metrics) and validated against scientific research agent benchmarks (e.g., AstaBench and MLGym). The work will also incorporate behavioral evaluation methods (e.g., Bloom-style targeted behaviors) to measure properties that matter in real research settings: citation faithfulness, source quality, appropriate tool selection, stopping criteria, and interaction quality.
Anticipated benefits for Tiptree include: (1) a reusable evaluation suite that functions as a regression test bed for product iteration, (2) higher-trust research agent performance for customers in academia and R&D, and (3) publishable methodology and results that strengthen Tiptree’s technical credibility and commercialization pipeline.

Optimal Quantum Circuit Partitioning via Tensor Network Methods

Quantum computers are expected to help solve important problems that are currently out of reach for classical computers, including challenges in chemistry, materials science, and optimization. However, today’s quantum devices are still limited in size and cannot run very large or complex programs on a single processor. This project aims to address this limitation by developing methods to divide large quantum circuits into smaller parts that can be run on multiple processors and then combined efficiently. By improving how quantum circuits are distributed and minimizing the communication required between devices, the project will help make better use of existing hardware.

Climate Change and Biodiversity Loss in Wadden Sea Salt Marshes – a UNESCO World Heritage Site

First-Principles Investigation of Non-Linear Optical Properties in Layered Materials

Développement de modules d’automatisation du contenu, de recherche sémantique et d’interrogation en langage naturel par RAG pour la plateforme SaaS spécialisée en gouvernance des données et de l’IA de Prodago

Ce projet vise à développer des outils d’intelligence artificielle pour automatiser la génération, la traduction et la recherche de contenus au sein d’une plateforme SaaS de gouvernance des données et de l’IA. L’objectif est d’enrichir les outils existants afin d’améliorer l’accessibilité bilingue et la découvrabilité de la base de connaissances de la plateforme

50th Anniversary of Economic, Social and Cultural Rights in Canada

This project will explore Canada’s record when it comes to its obligations under the International Covenant on Economic, Social and Cultural Rights over the past five decades, and what reforms could be implemented to strengthen accountability for the next fifty years.

The Freshman Shirt – Del-Coin Shirt Club – May 4, 2026 to August 21, 2026

This project aims to develop a data engine for ‘The Freshman Shirt’ digital platform. The intern will architect scalable infrastructure to support 100,000+ users , focusing on real-time data integrity and backend reliability for national collegiate rollouts.”