Joel Therrien

Data Scientist

Biography

I’m a Data Scientist / Statistician working for Landcor in New Westminster, B.C., where I’m working on improving their core product to automatically appraise homes. Alongside using industry-standard skills such as programming, databases, and SQL, I’m also developing, implementing, and experimenting with new machine learning models to improve said product’s accuracy.

In my personal life, I greatly enjoy self-hosting my own web services, including my own Gitea instance, Matrix homeserver, Tiny-Tiny RSS instance, personal website, and more. I also love working on the occasional personal programming project when I have time. In the past this has been writing mods for Minecraft. I also love reading fantasy novels and spending time with friends, or combining the two interests in Dungeons and Dragons.

Interests

Statistics
Self-hosting web services
Data Science
Programming

Education

MSc in Statistics, 2020

Simon Fraser University
BSc in Mathematics, 2015

University of British Columbia Okanagan

Skills

R

100%

SQL

100%

Statistics

90%

Java

90%

Linux

90%

Server Hosting

90%

Experience

Data Scientist

Landcor

Feb 2021 – Present New Westminster, BC

Designed and implemented a new automatic home valuation model, which significantly outperforms the preexisting legacy model in terms of accuracy, flexibility, and computational time. I built this in both Julia and R, implementing a random forest to act as an adaptive kernel function. The resulting model was then able to both provide highly accurate results while not sacrificing the interpretability that home assessments usually provide.
Mentored, guided, and taught a junior Data Scientist. Largely I helped her develop statistical intuition and learn programming best-practices, and encouraged her to develop a healthy work-life balance.
Created internal tooling to allow for a non-IT team member to retrieve property data. This was a simple website / Python back-end accessible through our internal VPN.
Wrote a tool to improve on the custom reports we provide clients. Instead of providing a static and frustrating format like a PDF, and instead of relying on a third-party BI tool (with all the headaches of authentication), I created a Python tool that pre-computes all the plot & table aggregations in the report and bundles it into a single HTML file (built using React). This report can then be merely emailed to clients, easily archived, and still interactive as the data can be filtered through drop-downs built onto the page.
Reverse-engineered legacy models & processes from an outdated language (SPSS command syntax) and rewrote them into R – simplifying the processes while also identifying bugs along the way that had been uncaught for years.
Built functional proof-of-concepts on new ways that customers can have self-service reports – examples being aggregate statistics on regions that the client themselves draw on a map (as opposed to pre-defined regions which we previously could only provide). These were built using React for the front-end and Python (FastAPI) for the back-end.

Graduate Data Peer

Simon Fraser University Library

Sep 2017 – Aug 2019 Burnaby, BC

Provided R software support to graduate students for their research.
Wrote and taught introductory R & Python workshops for graduate students.

Data Engineer

Grow Technologies

Nov 2015 – Aug 2017 Vancouver, BC

Wrote an ETL process to transfer data from a NoSQL database into a PostgreSQL database for analytics.
Setup and managed said PostgreSQL database and the VM it ran on.
Setup and managed a business intelligence tool (Looker) to assist my co-workers on their tasks.
Wrote a process to export loan performance data according to the credit bureau’s standards.
Assisted the software engineers as needed, including development work.

Student Research & Teaching Assistant

University of British Columbia Okanagan

Jan 2014 – May 2015 Kelowna, BC

Worked on a research project investigating how changes in land-use (as measured by satellite) could impact bird populations over time. I also was a TA for several courses in statistics.