I'm a third-year at the University of Waterloo studying statistics, drawn to big data pipelines, small local models, and shipping software that people actually use.
The quick version. Longer one lives on my LinkedIn.
I grew up taking things apart and (eventually) putting them back together. These days the "things" are data pipelines, ML models, and small pieces of software that sit between them.
I like to build products that won't just be generic, off-the-shelf — I'd
rather ship something creative that solves a problem people didn't know they had.
Bonus points if the data is inconveniently large and a simple
pandas.read_csv
won't cut it.
Outside of class I build for hackathons (four and counting), spend time at the gym bettering myself in ways school can't measure, and travel whenever I can — Japan is up next this fall.
A running log of co-ops, research, and teaching — in reverse chronological order.
Building new tools with senior mentorship to streamline internal data analysis, including parallelized workflows with MPI. Producing models, visualizations, and dashboards that situate Canada's position against global events in relation to energy, forestry, and mining commodities.
Cleaned and integrated external financial datasets in Excel, VBA, and Power BI with seasonal-adjustment transforms feeding the firm's microeconomic analysis. Lifted master-depot data quality by ~18% and presented case-study findings to PMs and advisors.
Automated legacy content migration in Python (pandas, Selenium, BeautifulSoup) with a 1:1 link-matching pass, cutting manual work by 95%+. Refactored 1,000+ Salesforce Knowledge articles, trimming customer lookup time ~40%.
Built ASP.NET/Blazor data-entry forms backed by SQL Server and CRUD APIs, automating 1,000+ leave and 5,000+ business requests annually. Added observability to the EMS, reducing MTTD by ~25%.
↳ Full resume on request — drop me a line.
Mostly hackathon builds — rough edges intact. Each was an excuse to learn something new.
Local LLMs, meet your editor.
Full-stack IDE with fine-tuned small models running entirely on-device. Explores local inference as a first-class citizen of dev tooling.
Sensors that notice before you do.
Wearable temperature tracking paired with software that learns your flare patterns. Hardware + software combo built in 36 hours.
Catching embers before they're fires.
Processes environmental sensor streams to surface early wildfire risk for first responders. Built for the "prevent the spark" prompt.
Syncing style with the forecast.
Personalized outfit recs from real-time weather, using a fine-tuned Llama 3.1 and Stable Diffusion. Yes, it actually picks the rain jacket.
The titles I keep returning to — the backbone of my data science roadmap.
Updated whenever I remember to. — last touched 4/19
↳ full roadmap: dsRoadMap