JAMES GLEAVE

Home: Welcome

BIO

I am an energetic 20 year-old athlete and computing student with an irrepressible desire to be challenged by complex problems and ideas. From a young age, I have been committed to learning, improving, and expanding my horizons, upholding these values in both my personal and professional endeavors. At age 14, I discovered programming and it instantly gripped me. My interest in mathematics came later, when I took a step back from the way I was taught and realized math was fluid and not as rigid as I originally thought. Machine learning fascinates me; the idea of teaching a computer to think is extraordinary and of limitless potential. My long-term goals align with my desire to harness artificial intelligence for creative solutions to critical issues like climate change, healthy longevity, and clean energy. I want to be a part of a future where machine learning will help humanity peel away the layers of complexity to enable great advancements.

Home: About Me

MY EXPERIENCE

work & volunteer experience

VANCOUVER PROSTATE CENTRE - STUDENT INTERNSHIP

July 2019 - August 2019

During my internship, I was responsible for creating and training an image segmentation software using deep learning.

HACKATHON - VOLUNTEER

Summer 2018

I volunteered at the UBC Great Hach. This was a tremendously inspiring experience.

PEDALHEADS - INSTRUCTOR

July 2015 - Sept 2015

PedalHeads is a camp where young children learn how to ride bikes. This job taught me a lot about patience, organization, and teamwork cooperation.

Home: Experience

EDUCATION

September 2018 - Present

QUEEN'S UNIVERSITY

Computer science undergrad with a specialization in artificial intelligence.

Summer, 2019

PIMS CRG SUMMER SCHOOL: DEEP LEARNING FOR COMPUTATIONAL MATHEMATICS

I took a summer school course in machine learning where I learned about the mathematics behind deep learning.

2013 - 2018

LORD BYNG SECONDARY

I graduated as an honour roll student. I played rugby from grades 8 to 12 and travelled with our team to New Zealand and Australia to compete. I took AP Computer sciences to get a headstart in university.

2006 - 2013

JULES QUESNEL ELEMENTARY

I attended a French immersion elementary school.

Home: Education

MY SKILLS

PROGRAMMING

I am experienced with Python, C, C#, Java, and Linux and I am capable of learning a new language quickly.

COLLABORATION AND WORK ETHIC

I am very driven to complete a task to the best of my ability. I am happy to collaborate, learn, and improve from those around me.

Home: Skills

MY PROJECTS

Here is a collection of projects I have enjoyed working on.

Home: Projects

PODIUM BOT

Puzzle Solving

This is one of my favourite projects. The following video is an example of relatively complex problem solving using reinforcement learning. The goal of the agent is to navigate a segmented square arena full of pillars holding blocks. The agent must find and press a button that spawns yet another pillar holding a block. This block, however, is gold and capturing this block is the only way the agent is rewarded.

LEARNING TO WALK

Baby Steps

The following videos are examples of gate optimization using reinforcement learning.
I give the agents an incentive to reach their target as quickly as possible. The faster the agent reaches its target, the higher its reward. Each attempt terminates if the body of the agent contacts the ground. Given enough time, the agents learn to run.

RGB LED MATRIX

Hardware

During the summer of 2019, I built a multifunctional RGB LED matrix that runs off of a Raspberry Pi. I created the software with modularity in mind, so it is very easy to add different features. I did not want to need to connect the Raspberry Pi every time I wanted to use the matrix, so I also created a wireless controller desktop app to connect to it over Wi-Fi! From audio visualizers to bubble sort, the matrix displays it in full beautiful colour. Unfortunately, recording LEDs never does it justice. The shutter on my camera is less than the update rate on the LEDs, so there are some visible issues in the video.

CYCLEAI

New Experience

Cycle AI is a mobile application that utilizes the power of machine learning to classify recyclables, organics, hazardous waste, as well as regular waste allowing users to properly dispose of their garbage. The process is very straightforward, simply take out your smartphone and snap a photo of your waste in CycleAI, and the app will inform you what it is you are disposing of, as well as how to dispose of it. CycleAI has a built-in achievement system, granting users rewards such for using the app consistently, scanning every category of recyclables, and reaching 100 total scans.

My team and I had initially created Cycle AI in 36 hours for QHacks, an annual hackathon hosted at Queen’s University. This entire project is written in Python, using Kivy for the front-end, and TensorFlow, Keras, and OpenCV for the back-end. Our original model utilized a standard convolutional neural network that could classify only one object per image. After receiving some constructive feedback from the hackathon judges, we have implemented an entirely new, more advanced neural network architecture that can isolate and mask several objects per image.

We ended up winning direct entry into the mayor's innovation challenge (https://innovation-challenge.cityofkingston.ca/), a pitch challenge hosted at Kingston town hall.