☆ Automated Third-Party Library Upgrades

▼

☆ CI/CD Integration and Efficiency Gains

▼

☆ Security and Compliance

▼

☆ Contributed to enhancing workflow efficiency in FrostEd, the C# editor for the game engine

▼

☆ Enhanced expression debugging for 10+ game teams by developing instance selection feature

▼

☆ Reduced Undefined Behaviour bugs for cross-platform PS5 builds by fixing member calls on unaligned addresses

▼

☆ TA for CPSC 213 - Introduction to Computer Systems

▼

☆ TA for CPSC 110 - Computation, Programs, Programming

▼

☆ Developer on game: Cybercision, a Cybernetic Surgery Simulator

▼

☆ Designed and developed a shooter ”chip debugging” level

▼

☆ Won ”Audience Choice” and ”Best Audio” at End of Year Showcase

▼

Technologies Used: JavaScript, ExpressJS, Firebase, OpenAI Turbo Vision Model

Winner of the LLM enhancer category at Hack the North 2024, a $2000 prize. A camera records you as you go about your day to day life, storing "comic book strip" panels containing images and context of what you're doing as you go about your life. When you want to remember something you can ask out loud, and it'll use Open AI's API to search through its "memories" to bring up the location, time, and your action when you lost it. This can help with knowing where you placed your keys, if you locked your door/garage, and other day to day life. The React-based UI interface records using your webcam, screenshotting every second, and stopping at the 9 second mark before creating a 3x3 comic image.

Technologies Used: React, Node.js, Tensorflow, Blazepose, ExpressJS, Axios, Figma, GPT-3

First Place against 665 participants at NwHacks 2023, Western Canada's Largest Hackathon. Yudo, an exercise web app, leverages TensorFlow AI and a custom-developed exercise detection algorithm to assist users in improving their exercise form. Users select a workout, and Yudo provides a demo video, scoring them based on form accuracy. The app utilizes BlazePose and JSON to stream video data into TensorFlow, creating dynamic 2D visualizations of exercise forms. OpenAI integration with GPT-3 provides personalized feedback on plank form through an ExpressJS back-end, helping users identify and correct form discrepancies. The front end was designed with Figma and Procreate and implemented with React.

Technologies Used: C++

Developed an algorithm utilizing Breadth-First Search (BFS) traversal and distance encoding to generate a structured path simulating the ideal base location. This is based on the idea that opposing teams have bases strategically placed at the furthest distance apart on in-game maps. A custom deque was created based on a vector, maintaining constant time operations by automatically resizing upon element size changes. The algorithm calculates distances for every pixel along the map's paths, determining the base's location as the longest shortest path from a specified starting point. The resulting path is then visually rendered on the image, providing players with a strategic perspective.

Technologies Used: Python, Firmata, Lilypad, OpenCV, PyFirmata, Solidworks

Crafted a 3D printed robotic arm that utilizes face tracking to move a spoon toward your mouth, streamlining the process of delivering food directly to the consumer. The design incorporates the ability to scoop sustenance from a bowl, ascend towards the user with an open mouth, and retract upon closure, automating the feeding process based on face tracking. This robot arm features custom-modded 3D printed components and is powered by an Arduino Uno, equipped with 6 SG90s servos, and supported by 2 power supply modules. MediaPipe FaceMesh and OpenCV were used for facetracking, capturing video from a USB external webcam mounted on the arm, drawing landmarks onto the video, and extracting facial landmark positions for each frame. Leveraging the Jacobian pseudo-inverse method in NumPy, the face tracking data is utilized to calculate reverse kinematics, determining the necessary angles for arm movement. The calculated data is then transmitted to the Arduino board, loaded with the Firmata protocol, enabling real-time servo control using PyFirmata.

Technologies Used: React, ExpressJS, HTML, CSS, GPT-3

Developed a personalized shopping experience for Aritiza as a part of Girlcode 2023, allowing users to click on a preferred clothing item and access an extension featuring a lookbook of stylistically similar pieces. Each lookbook is tailored to specific aesthetics, such as "cozy" or "athletic," generated by backend that dynamically fetches AI-generated style recommendations based on keywords associated with the main clothing article. The webpage, made in React, integrates with a dataset of 40 Aritzia clothing articles. Users can regenerate aesthetics multiple times and save preferred looks to their personal lookbook, enhancing the overall customization and user-centric shopping journey.

Technologies Used: Java, JUnit

A game centered around feeding a Capybara named Kiwi. Users can choose from a variety of food items, including watermelon, library books, and fresh linen, to satisfy Kiwi's cravings. The game allows users to build a feeding history list, observe specific reactions based on the food item chosen, and save or load their feeding history. Additionally, users can view the calorie count of the food items they've fed Kiwi. The user interface, located at the top of the page, displays Kiwi's avatar and mood, while interactions are facilitated through buttons such as 'Feed Kiwi,' 'Save History,' and 'Load History.'