Hayson Cheung

My Interests

My technical interests span machine learning, AI, robotics, and control systems. I love leveraging software and emerging technologies to enhance system performance and solve practical problems.

Projects

• Krackle

  Developed a real-time emotion detection system using CNNs and Eigenface algorithms to track user reactions during video interactions, enhancing user engagement.

• NEAT Pong Bot

  Created a high-performance Pong bot using NeuroEvolution of Augmenting Topologies (NEAT), exploring deep learning techniques like CNNs and Q-learning for improved decision-making.

• MapMatch

  Designed an AWS-based algorithm to analyze roommate descriptions and compute compatibility through semantic embeddings, efficiently matching roommates with similar habits.

• PHY-Pendulum

  Developed an object detection system using OpenCV to analyze pendulum motion with center origin detection and dynamic graph generation for detailed analysis.

• AWS Hacks

  • Built a backend algorithm that analyzes roommate descriptions by extracting keywords, transforming them into semantic vectors, and assessing compatibility using AWS services.
  • Enhanced matching efficiency and accuracy for streamlined pairing; featured for potential deployment at UofT.

• Smash Bros. RL Agent

  Developed a reinforcement learning agent for a Smash Bros.-style platformer game using rPPO, NEAT, and Stable-Baselines3. Integrated recurrent networks with RecurrentPPO to improve decision-making and evolve optimal network topologies.

• Node-Based Machine Learning Model Builder

  Developed a Django app that generates machine learning scripts based on user-defined pipelines. Includes builders for PyTorch and statistical models to streamline custom model development.

Related Experiences

• University of Toronto Aerospace Team - Control Systems

  Developed MATLAB and Python-based control systems for the FINCH satellite mission, contributing to precise altitude planning and orientation control in space systems.

• Yannes Solution Ltd.

  Streamlined project documentation and managed the installation of over 500 CCTV cameras for Hong Kong Metro’s East Rail Line, optimizing image processing and installation pipelines, resulting in faster project delivery and cost reduction.

• VEX Robotics

  Led a team in designing and building competitive robots for VEX Robotics tournaments, focusing on performance improvements through iterative testing and design optimization.

• University of Toronto Machine Intelligence Student Team (UTMIST)

  Led academic paper-reading sessions on LSTMs, Encoder-Decoder models, and Transformers, and developed an LSTM Encoder-Decoder model for Neural Machine Translation using TensorFlow, improving team knowledge and technical skills.

• Coding Club - Ridley College

  Instructed dynamic programming, linear programming, and hashing algorithms, mentoring 40 novice students and helping 7 qualify for the Canadian Computing Competition. Also organized STEM events for 150 junior students, sparking early interest in technology.

Skills

ESROP-Global Scholarship

Awarded a $3,500 scholarship for a research position in Thailand at KMUTT, where I focus on machine intelligence, neural networks, and robotics.

Resume

For more details on my experience, education, and skills, please view my full resume.