Selected Work

Do You Think GPT Will Be Correct?

We designed an intervention with metacognitive activities to improve student's generative AI literacy.

UofT Summer Undergraduate Research Showcase

HCI Research

Receiptify App - Receipt Management Made Easier

Backed by research, we designed an application to help small business owners and individuals manage their receipts better.

Product Design Concept

Cyanotype Digital Studio

An app where users can create their own digital cyanotypes using their photos.

Design Engineering Shipped

The Interplay of Paralinguistic Cues and Expressions of Certainty in Human-Human and Human-Agent Interactions

We provided design recommendations for presenting uncertainty using paralinguistic cues in voice agents, backed by research.

ACM TOCHI 2026

HCI Research Applied Machine Learning Shipped

Nourpact (2nd place winner - college)

A web app with the goal of reducing food waste by locating nearby food rescues and finding recipes using given food items. Made using React, CSS, and SerpAPI. Hosted on Netlify and created high-fidelity wireframes on Figma. Won 2nd place (college) and best domain name out of 96 projects (with 430+ participants) at SimpliHacks 2.0.

Web Development Shipped

Street Whisper App

A command-line app that transcribes and translates audio files across different languages for research projects. Uses OpenAI's Whisper API and pyannote.

Software Development Shipped

Characteristics of Top Songs Has Changed from Pandemic Brain: An analysis of songs on Billboard's Year-End Hot 100 singles list (2014 to 2023)

Music often reflects the current climate of society and there is a growing interest in how hit songs as seen on Billboard's Year-End Hot 100 singles has changed after the COVID-19 pandemic. This paper looks at music characteristics such as tempo, song duration, loudness, and modality of songs from Billboard's Year-End Hot 100 singles list from 2014 to 2023 to reveal patterns and relationships to explain the difference between top songs before 2020 and 2020 onwards.

Telling Stories with Data: With Applications in R Book

Data Science Bayesian Statistics Shipped

Recommended or Not?: Applying Supervised Machine Learning Models to Understand UofT Course Evaluations

Machine learning paper using models such as a multilayer perceptron neural network with Adam optimizer, logistic regression, linear regression, decision tree, and Gaussian Discriminant Analysis, as well as a majority voting ensemble to understand factors influencing whether or not a UofT course gets recommended or not by students.

Applied Machine Learning

Steering Moral Sycophancy in Language Models

Using activation steering to decrease sycophantic behaviour in language models Llama-3.2-1B-Instruct and Qwen2.5-1.5B-Instruct.

Toronto AI Safety Initiative (TAISI) Capstone

AI Safety