AI & Data Science
AI & Data Science
AI and Data Science projects explore how intelligent algorithms can analyze data, make predictions, and support decision-making in real-world contexts. Students apply methods such as machine learning, natural language processing, and predictive analytics to develop solutions that are innovative, practical, and responsible.
DFFRNT AI Assistant: Air-Gapped Private AI Knowledge & Content Assistant
DFFRNT, a strategic design consultancy, needed the productivity of a modern AI assistant without ever sending confidential client data to the cloud. The group built a fully air-gapped, open-source RAG system that lets any consultant ask natural-language questions and get cited, trustworthy answers from the firm's own document base — in seconds instead of tens of minutes.
Aegis is an enterprise governance layer for agentic AI systems, built in partnership with Kinaxis. It sits above existing AI tools such as Copilot, Claude, ChatGPT and enforces organizational policies at every agent action in real time. The system is grounded in several frameworks: OWASP Top 10 for LLM Applications, OWASP Agentic-AI threats, MITRE ATLAS and the NIST AI Risk Management Framework.
Every model call, tool call, memory read, and write-back passes through a single Policy Decision Point that enforces four levels of governance simultaneously: org-level invariants nobody can override, team-level policies, role-based permissions, and individual preferences. This is the L4 Hierarchical Values Cascade. Aegis's novel research contribution resolves at request time in under 100ms, with every decision recorded in a tamper-evident, hash-chained audit ledger.
DataLens Ottawa is an automated system that evaluates the quality of Open Ottawa datasets, detects key data issues, and presents clear results through a dashboard. It can give feedback from five dimensions, including accessibility, usability, freshness, metadata, and completeness. Then it can provide the relevant automatic suggestions for users on which kind of application they should use, and which is not suitable for the related datasets.
InvisiFall | ML-Enhanced Radar Fall Detection System for Retirement Homes
The InvisiFall project develops a radar-based fall detection system for elderly care, utilizing FMCW mmWave radar technology to track movements and detect falls in real-time. Unlike traditional wearable solutions, this system is non-intrusive, preserving privacy while ensuring safety. The system uses machine learning algorithms to analyze radar data, trigger alerts, and notify caregivers about falls. It is designed to work in various residential environments, including nursing homes, providing enhanced protection for elderly individuals, particularly those with cognitive impairments. By eliminating the need for wearable devices, InvisiFall addresses the discomfort and risk of falls without compromising user privacy. The system aims to improve response time, reduce fall-related injuries, and increase the quality of care for elderly residents, ultimately making it a valuable tool in senior healthcare management.
E-Hospital | AI-Powered Healthcare Solution for Smarter, Safer Patient Care
The E-Hospital platform redesign focuses on improving the user experience for both doctors and patients by addressing usability issues and enhancing workflow efficiency. For doctors, the platform’s dashboard was redesigned to better reflect clinical routines and minimize cognitive overload. Key features like the patient list, patient overview, and encounter sections were optimized for quicker access to essential information. For patients, the platform simplifies tasks such as booking appointments, reviewing lab results, and requesting prescription refills. The redesign incorporates user-centered design principles to ensure that both patient and doctor interactions are intuitive, efficient, and aligned with their needs. AI features were integrated to enhance decision-making for doctors, while streamlined navigation paths improve task recognition for patients. Extensive usability testing and iterative design ensured the platform’s enhancements meet the expectations of healthcare professionals and patients, ultimately leading to a more accessible and effective E-Hospital platform.
Cell Free Layer Detection using AI
The "Automatic CFL Detection Using AI" project develops a sophisticated application for automatic measurement of Cell-Free Layer (CFL) thickness within blood vessels, which is crucial for understanding blood flow dynamics and the effects of microcirculation on blood viscosity and gas/nutrient exchange. The system leverages Convolutional Neural Networks (CNN) to fully automate the segmentation of CFL in medical images. The application is designed to be fast, user-friendly, and adaptable to varying image quality and sizes, ensuring accuracy across diverse datasets. Key metrics include prediction accuracy, application size, response time, and user feedback, with targets aimed at improving performance and usability. The system incorporates edge detection techniques and advanced semantic image segmentation to provide accurate results, while also allowing for batch processing of images and videos. The solution is designed with ethical considerations around data privacy and inclusivity in mind, ensuring the system is accessible and compliant with medical standards.
PDF INSIGHT ENGINE
The PDF Insight Engine is an AI-powered, privacy-first solution designed to transform how users interact with unstructured PDF documents. Built for industries such as healthcare, finance, legal services, and research, the system enables real-time extraction, summarization, and querying of long and complex PDFs through a conversational, chat-based interface. Using advanced Natural Language Processing (NLP) and Large Language Models (LLMs), the engine accurately parses text, tables, and document structures while allowing users to ask both simple and complex questions. A strong emphasis is placed on data security and ethics, with features such as AES-256 and RSA encryption, secure user authentication, session management using JWT, and non-persistent storage to protect sensitive information. Deployed using a scalable serverless architecture, the PDF Insight Engine significantly reduces manual document analysis time, improves decision-making efficiency, and ensures compliance with data protection standards, offering a secure, intuitive, and intelligent document analysis experience.