Current
My ongoing research, which spans both my MASc graduate studies and industry partnerships with Nissan and Apple, touches on two main factors impeding the adoption of emerging AI-based imaging solutions: (1) providing per-pixel annotations for imaging tasks is very expensive and labour-intensive; (2) satisfying small margins of error often requires designing highly specialized models that are brittle to small variations in input data (i.e., domain shifts).
May
2023
MSCI 436 Decision Support Systems. This course provides an introduction to the design and implementation of AI-based decision support systems for engineering and business applications. My responsibilities included tutoring, consultations with students, marking and creating examinations and assignments.
Jan
2022
Worked under UWaterloo's Vision and Image Processing (VIP) Lab focusing on egocentric 3D human pose estimation research. This culminated in co-authoring 1) a CVPR 2022 workshop conference paper and 2) a KDD 2023 conference publication. My efforts consisted of: reviewing and presenting background research on papers; collaborating with team members to generate novel ideas in weekly brainstorming sessions; implementing validation and evaluation modules for the PyTorch-based model prototypes.
May
2021
Prototyped recurrent neural network architectures for wearable and computer vision applications. Primary focus was on human activity recognition from human electrocardiogram signals.
Jun
2020
Identified and implemented point-cloud surface feature descriptors for LiDAR-based human/vehicle recognition with C++11, PCL and OpenCV. Prototyped SVMs and Neural Networks to improve the precision and recall of the Helius Smart Lidar. Researched and prototyped LiDAR-based ego-motion vehicle estimation models on the Automotive Perception Team.
Sep
2019
Developed multi-channel J2K encoding/decoding products for Amazon and ESPNs streaming services using C/C++. Developed data corruption detection modules for video and audio ST2110-20/30 IP packets in python. Collaborated with product specialists to perform endto-end integration tests for the 570J2K-HW-X19 encoder.
Jan
2019
Developed navigation wizards to integrate Vena’s FP&A platform with external data sources using React and Redux. Debugged the Java 8 back-end to diagnose issues with MongoDB and SQL Server transactions. Composed technical documentation for the research and testing stages of Vena's feature development cycle.
May
2018
Redesigned RESTful APIs purposed with on-boarding insurance brokers; saved +$20k/ year worth of labor fees. Automated application build and deployments with Jenkins to improve production process efficiency. Developed and deployed RESTful APIs on Azure and AWS cloud services.