Selected work

Projects

A selection of systems, experiments, and products I have built with collaborators and on my own.

MetaAI ASRBleu Pipeline
ML infrastructure

MetaAI ASRBleu Pipeline

Engineered a Hydra-configured evaluation pipeline for benchmarking 25+ multilingual speech-translation models at scale, reducing turnaround time and enabling repeatable Meta AI research workflows.

View details ↗
Chip
Natural language processing

Chip

A multifunctional Discord bot to serve my student community at UWindsor, featuring an interactive RAG-based chatting feature providing immediate information access to students as well as a fine-tuned BERT-based automatic moderation system to flag harmful content.

View details ↗
The House Where No One's Living
Reinforcement learning

The House Where No One's Living

For my undergraduate capstone project, I created an immersive fantasy themed RPG featuring adaptive enemy behaviour with reinforcement learning powered by Unity MLAgents and a SAC model trained through thousands of rounds of self play.

View details ↗
64squares
Reinforcement learning

64squares

Trained a chess engine via self-play and deep reinforcement learning to explore efficient policy learning and strategic decision-making under constrained hardware.

View details ↗
PongOS
Systems programming

PongOS

A minimal operating system that boots and runs Pong directly on the CPU, handling low-level tasks and device drivers in C and x86 assembly in 32-bit mode.

View details ↗
Autoquest
Hack Dearborn 2023 · Second place out of 113 participants

Autoquest

Created an accessible vehicle-selection web app for the Ford challenge, combining user-centered design with an interactive decision flow to simplify a complex purchase process.

View details ↗
Foodbox
McHacks 2023 · Second place out of 401 participants, TELUS Sustainability Award

Foodbox

An app that partners with grocery stores to deliver surplus food to people in need, addressing food waste while promoting environmental sustainability.

View details ↗