I am a first year ECE Master's student at the University of Toronto supervised by Ashish Khisti. My interests include information theory, statistical learning theory, probabilistic models and other applications of probability or optimization. I am especially interested in applying these to better understand machine learning.
I obtained my undergraduate degree in Electrical Engineering at McMaster University with minors in Mathematics and Computer Science. There, I worked on a project on classifier trading supervised by Hassan Ashtiani.
A Rover that finds and disposes of cans without any human intervention. The core logic runs on a Raspberry Pi and communicates sensor/camera data to an external server running an object detection model. Capstone project completed in a team of 4; I worked on the object detection.
Visualize the prerequisite requirements of McMaster's courses. Text descriptions were taken from McMaster's academic calendar and parsed into a tree modelling the logical dependencies between courses. On the browser end, these trees are recursively merged into a graph to generate the full dependence structure.
An analysis of kanji usage frequency in Japanese-speaking subreddits. Data collected with the reddit API, managed with an sqlite3 database and analysed using numpy/pandas. Multiple visualizations generated using d3.js and HTML/CSS.
An optimization-based approach to multilateration with an unknown number of targets. Multilateration is the technique of using only distance measurements from stations to determine the location of a target, with no information about the direction from which this distance was measured.
Implemetation of "A Linear Programming Approach for Optimal Contrast Tone Mapping" in MATLAB. Enhances the contrast of color images under poor lighting using a contrast mapping technique defined by linear objective and constraints. Project for Comp Eng 4TN4 in a group of 2.
A hackathon project aimed at finding out positivity/negativity trends around a geographical location. Collects tweets made close to a point in space and performs sentiment analysis by assigning a positivity score using a Naïve Bayes model. Hackathon project completed in a group of 3; I worked on training the sentiment analysis.
I'll continually add to this as I think up of things.