Kevin Ni

I’m completing my BASc in Engineering Science at the University of Toronto, majoring in Robotics Engineering with double minors in Engineering Business and Artificial Intelligence.

My coursework has built a strong foundation across AI and finance. On the AI side, I’ve studied Introduction to Artificial Intelligence, Learning from Data, and Applied Fundamentals of Deep Learning. On the finance side, I’ve taken Financial Engineering, Accounting and Finance, and Economic Analysis & Decision Making.

In 2025, I was awarded the Jeffrey Skoll Engineering Business Scholarship, which recognizes EngSci students pursuing the Engineering Business minor who demonstrate strong potential in business and entrepreneurship.

As part of my 16-month co-op term, I joined Smith + Andersen with no prior background in building systems. Despite that, I quickly immersed myself in the role — producing drawings in AutoCAD and Revit, coordinating with architects and contractors, and conducting site surveys to ensure designs were implemented correctly.

On the job I learned transformer and panel sizing, reviewed budgets and quotations, and developed a strong sense of how design decisions impact costs, efficiency, and client outcomes. The experience is a clear example of my ability to learn quickly, adapt to new fields, and provide value to the company in a fast and meaningful way.

This was my first internship in the robotics field, and it gave me a solid introduction to what it’s like working at a startup — fast-moving and full of figuring things out as you go. My main task at Kevares was building a streaming pipeline using Streamlabs, which let us send live camera footage directly to a browser. Near the end of my term, I started planning out a system to automatically detect sidewalk damage from the live feed. I didn’t get to finish it before the internship wrapped up, but the experience got me excited about using robotics and computer vision to solve real-world problems.

I was thrown into code early, and while that was intimidating at first, it ended up being a crash course in learning on the fly and writing software that actually mattered. I also started to see how robotics, AI, and web tech could intersect in unexpected ways — something that continues to shape how I think about my career in fintech and automation.

Pushing carts all day in the heat wasn’t the most glamorous job, but it was real work — tough, physical, and surprisingly rewarding. I learned how to power through long shifts, stay positive even when it was 35°C and humid, thunderstorming or even blizzarding. It truly shifted my perspective and made me appreciate what it means to work hard.

The best part was the team - family that I still keep in touch with to this day. We looked out for each other, especially during the crazy rushes or when the weather turned bad. I made some great friends there, and it showed me how much easier hard work feels when you’re in it together.

My first job taught me the fundamentals of customer service and working in a fast-paced environment. I learned how to multitask effectively while maintaining quality, whether I was taking orders, making drinks, or preparing food during busy rushes.

Working in a small team environment taught me how important communication and reliability are. I gained experience handling cash, managing inventory, and maintaining food safety standards — skills that built a strong foundation for my work ethic.

For my final year, I’m working with Professor Roy Kwon on a thesis that bridges my two big interests: artificial intelligence and finance. The project is still unfolding, but it’s focused on how machine learning can help tackle real problems in financial markets — things like high-frequency trading, predicting trends, and risk management. It’s the kind of work that pulls together everything I’ve been building toward in engineering and business.

In this course project, my team and I set out to push the limits of audio generation. We experimented with Test-Time Training (TTT) and a Diffusion Transformer to make long audio clips sound more natural — less repetitive, more coherent. To get there, I had to wrangle over 100,000 audio samples with Python and FFmpeg, cleaning and standardizing everything until it was usable.

The coolest part was seeing our custom model outperform a baseline we compared against, especially under tight memory limits. It was one of those projects where the results felt both technical and creative, which is exactly why I loved it.

These labs were where I really started to get my hands dirty with machine learning. I built models like k-Nearest Neighbors for regression and classification, tested out linear and logistic regression with different optimizers, and even trained a neural network on MNIST digits.

It was equal parts trial-and-error and “aha!” moments — especially when I hit 95% accuracy after tweaking and tuning. That feeling of getting something to finally click kept me hooked.

This was my first real dive into finance. We analyzed Caterpillar Inc. and Deere & Co., combing through financial statements and running ratio analyses to see how they stacked up. On top of that, we built a discounted cash flow model for a $120M nuclear energy project — crunching numbers on things like NPV, IRR, and tax shields.

It was intense, but it showed me how data and models could directly translate into investment decisions. That mix of numbers and strategy is what pushed me closer to fintech.