👋 I am an AI and computer vision professional with an M.Sc. in Electrical and Computer Engineering from the Computational Intelligence Laboratory at the University of Manitoba.
👩🏻💻 My work combines computer vision, deep learning, AI product development, technical project management, and product management. I have experience building and leading AI-driven products across areas such as creator attribution, healthcare technology, industrial IoT, and cloud-based platforms.
🌱 My technical interests include computer vision, AI/ML systems, data storytelling, AI product development, technical project management, and responsible technology that creates meaningful social impact.
💻 I enjoy working at the intersection of research, product, and execution. I am especially interested in building AI-powered solutions that solve real-world problems, support creativity, and strengthen society and culture.
👀 In my spare time, you can catch me watching sci-fi movies and reading spiritual and self-development books.
Feel free to connect with me!
Temporal Multiple Moving Objects Recognition Using Shape-Based Descriptor Matching | Thesis Document | Demo
This is my thesis project for graduate studies (M.Sc.), which I did at the University of Manitoba's Computational Intelligent Lab. The algorithm extends the approach of extracting moving objects in complex backgrounds in the temporal domain. Shape descriptors in 4 dimensions are then estimated and utilized for shape matching. The real-world application is multiple object recognition by using only 1-training sample per class and then a shape descriptor matrix to assess the proximity between various moving objects and establish their descriptive nearness (similarity) for recognition.
Real Time Single Object Detection and Tracking with Computational Geometry | Demo
This project describes an adaptive learning approach to detect, segment, measure, and track objects in outdoors, such as vehicles, pedestrians, etc. In this project, the object is detected using computational geometry, topology, and engineering physics rather than using neural networks. Therefore, only 1 training sample per class is required, rather than loads of visual data.
Real Time Multiple Moving Vehicle Detection and Segmenation with Computational Geometry | Demo
This project describes an adaptive learning approach to detect and segment objects that are vehicles on the road by applying computation geometry, topology, and engineering physics rather than using neural networks. Gaussian Mixture Model is employed for detecting moving foreground, and tesselation is used for segmentation.
Deep Learning Algorithm of a Snapshot Mobile App for DVD Finder | Project Report with Results
This project describes a deep learning algorithm for a mobile photo app where you take a picture of a DVD, and the app tells you all the information about it. The output should be to find out which movie the DVD cover belongs to. As each class of a DVD cover has a single training data instance, data augmentation was performed to increase training samples. The approach to classifying the DVD cover is based on the Siamese Neural Network that determines if the two inputs are different or similar.
Image Classification of CIFAR-10 Dataset
The classification of the CIFAR-10 dataset of 50,000 training images has been improved in this project. This is achieved by the logistic regression model-based convolutional neural network with Keras API of TensorFlow for object recognition.
Predictive Modeling For Canada COVID-19 Vaccinations | Project Report with Results
This project describes a statistical, descriptive, and predictive model using R and Python data analysis and data visualization tools. The model aims to forecast exactly when every Canadian will be completely vaccinated.
Idometrics
Idometrics is an AI-powered platform built to support dance creators with attribution, recognition, and creative insights. The platform uses technologies such as computer vision, pose detection, similarity analysis, blockchain-based registration, context learning and cloud infrastructure to help preserve creator identity, cultural context, and originality in dance.
Current Status: MVP launched with early users, ongoing product validation, and grant-supported research collaboration with McMaster, Universal Production Music and the Hamilton Conservatory for the Arts Society.
Healthcare Matrix | MVP Demo
Healthcare Matrix is a med-tech hardware startup that manufactures an interactive AI-driven standalone unit. The unit automates primary-level medical inspection and medication in third-world countries and remote areas. It also provides a basic medical report and releases over-the-counter medicines until the doctor checks the patient for further inspection. As a founder of this startup, I led and was involved in the ideation stage and worked through research, coding of FPGA myRIO embedded board, and hardware development along with the integration for all of its versions. Furthermore, I acted as a Chief Executive Officer and took business development responsibilities to improve the product and its marketing.(Patent Protected)