cv

Computer Vision Intern

NVIDIA

Incorporated synthetic data for training lane detection models.

• Developed data pipeline and experimented with synthetic data incorporation.
• Resolved critical failure scenarios blocking PathNetV4 release.
• Developed tools to better visualize and identify challenging scenarios.

Computer Vision Intern

NVIDIA

Time series light signal detection for autonomous vehicles.

• Detection of blinking light signals.
• Defined labelling guidelines and potential model architectures.
• Trained proof of concept classification network using public data.

Computer Vision Intern

Miovision

Traffic data analysis with computer vision.

• Led project to introduce active learning techniques to data ingest pipeline.
• Optimized selection of images for labeling from large unlabelled pool.
• 42% improvement in mean average precision between models.

Computer Vision Intern

Synapse Technology

Developed and analyzed CNN models for detecting threats in security x-ray scans.

• Developed fine grain rotational data augmentation method.
• Significantly improved model performance in underrepresented classes.

Data Scientist

Praemo

Used LSTM to detect anomalies in time series vibration data and predict machine failure in industrial robots.

Robotics Software Developer

ESI

Robotic navigation using reinforcement learning and IR sensors.

• 95% success rate in simulation and 85% success rate on physical robot.

PhD

University of Waterloo

Vision and Image Processing Lab, supervised by Dr. Alexander Wong

MASc

University of Waterloo

Vision and Image Processing Lab, supervised by Dr. Alexander Wong

BASc

University of Waterloo

Mechatronics Engineering

Finalist at 2022 IEEE International Conference on Automation Science and Engineering

IEEE

Decoding Diffusion: A Scalable Framework for Unsupervised Analysis of Latent Space Biases and Representations Using Natural Language Prompts

Neurips Safe Generative AI Workshop 2024

Measuring biases and visualizing concept clusters through sampling from the H-Space in diffusion models

Explaining Explainability: Towards Deeper Actionable Insights into Deep Learning through Second-order Explainability

CVPR 2023, XAI4CV workshop

A study on deep learning explainability presented at the CVPR 2023 XAI4CV workshop.

ShapeShift: Superquadric-based Object Pose Estimation for Robotic Grasping

CVPR 2023, WICV workshop

A paper on superquadric-based object pose estimation for robotic grasping presented at the CVPR 2023 WICV workshop.

COVID-Net US-X: Enhanced Deep Neural Network for Detection of COVID-19 Patient Cases from Convex Ultrasound Imaging Through Extended Linear-Convex Ultrasound Augmentation Learning

CVPR 2022, WICV workshop

Development of an enhanced deep neural network for COVID-19 detection from ultrasound imaging, presented at the CVPR 2022 WICV workshop.

MetaGraspNet: A Large-Scale Benchmark Dataset for Vision-driven Robotic Grasping via Physics-based Metaverse Synthesis

2022 IEEE International Conference on Automation Science and Engineering

Research on a large-scale benchmark dataset for robotic grasping presented at the 2022 IEEE International Conference on Automation Science and Engineering.

AUTOREAD