Mapping Canada’s wetlands using satellite data.

I built WetMaps with three teammates for my engineering capstone at Queen’s. The tool identifies wetlands and shows them on a map, helping update wetland maps without having to survey every area by hand.

How it works

The AI model learned what different wetland types look like in satellite data using existing wetland maps as examples. It then predicted where bogs, fens, marshes, swamps, and shallow water were located in other areas.

The first version produced patchy maps. Looking at the surrounding area, rather than each point on its own, helped create clearer boundaries. Users could explore the finished maps in Google Earth Engine, an online mapping tool.

My contribution

• Built the cloud-based process for handling large satellite files and turning the computer’s predictions into maps.
• Automated tests to make experiments consistent and repeatable.
• Checked how reliably the system worked with different data and across parts of Canada.

Results

The system averaged 87% accuracy across five rounds of testing, above the project’s 80% target.

A comparison of maps from 2017 and 2024 highlighted where Ontario’s wetlands appeared to grow or shrink. Some differences may come from changes in the satellite data rather than real changes on the ground, so these results need further checking.

Tools

Python, TensorFlow/Keras, Scikit-learn, Google Earth Engine, Google Cloud Storage, Google Colab, Docker.

A place for students to share and explore course and professor reviews.

I built and deployed a review platform using MongoDB, Express, React, and Node.js, with hosting on Microsoft Azure.

My contribution

• Developed the application for students to publish and browse ratings.
• Set up GitHub Actions to automate testing and deployment.
• Implemented JWT authentication, input validation, and profanity filters.

Tools

React, Express, Node.js, MongoDB, Microsoft Azure, GitHub Actions.

An expense tracker with receipt uploads and personalised spending summaries.

I built ReceiptMe with three friends during a 48-hour hackathon. The idea came from my frustration with paper receipts: I saw them as outdated, easy to lose, and a source of unnecessary, non-recyclable waste. I wanted to turn that everyday throwaway record into something useful.

How it works

ReceiptMe brings receipts and expenses into one place, helping users keep track of purchases and understand where their money goes. Personalised 30-day spending summaries turn those records into insights users can act on.

My contribution

• Built receipt-upload and expense-tracking features.
• Integrated the Gemini API to produce personalised 30-day summaries from users’ spending data.
• Implemented Google OAuth and spending comparisons by age, location, or income.

Tools

React, Express, Node.js, MongoDB, Gemini API, Google OAuth.

Engineering design experience on the manufacturing floor.

During my year at Direct Coil in Kingston, Ontario, I worked on drawings for heat-transfer equipment, including fluid coolers and condensers. Rotating through departments helped me understand the products, manufacturing processes, and practical constraints behind each design.

Highly customized jobs with tight deadlines were a constant challenge. I kept in close contact with customers and the manufacturing department to work through design questions, communicate changes, and keep everyone clear on what each job needed.

My work

• Created 2D and 3D parts drawings in AutoCAD.
• Worked with manufacturing teams to resolve production challenges and improve manufacturability.
• Communicated with customers to resolve design conflicts and balance performance with practicality.
• Contributed to revisions of design standards and managed time-sensitive jobs.

Areas of experience

AutoCAD, heat-transfer equipment, manufacturing processes, design standards, customer communication.