Sean Conley Email & Phone Number

Amsterdam, North Holland, Netherlands

Vancouver, British Columbia, Canada

Penn Foster is an online school focused on middle-skills training that offers diploma and degree programs to students. As a data scientist there, my work involved:• Designing, running, and analyzing A/B tests to measure changes to the user enrolment flow with respect to company KPIs.• Building statistical models of learner progression to better understand student attrition and identify the most relevant KPIs for the company to focus on.• Supporting the transition to Databricks and a Spark-based workflow, for example by expanding an internal Python package to allow Databricks jobs to be defined in code and stored in a version controlled repository.• Continuing all Sokanu-related work responsibilities, as highlighted below.

Vancouver, Canada Area

Sokanu is a company offering career and education guidance through the CareerExplorer platform. As a data scientist there, my work involved:• Designing, training, and monitoring career and degree recommendation models built using PyTorch.• Building, deploying, and maintaining microservices to serve model predictions to users using Django, Terraform, and CI/CD best practices.• Using a Bayesian hierarchical modelling approach to combine crowd-sourced career and education statistics, such as salary and tuition, with industry and government data to present information on a wide range of careers and degrees for various user demographics.• Translating business problems into data science problems, for example by combining NLP techniques with a graph-based algorithm to suggest related content and better optimize internal linking across the website for SEO.• Managing data infrastructure on AWS, such as PostgreSQL databases and a data lake, using Terraform Cloud.

Vancouver, Canada Area

A three month project focused on improving performance of a convolutional neural network for text classification, developed with MXNet.Developed interactive tools to visualise and interpret decision boundaries and modify target classes to refine and improve the model's performance.Focused on online learning to take re-annotated or new data and continue training a currently employed model.

Hamilton, On

Created and delivered tutorials to students, from class sizes of 10-15 students to over 100 students.Assisted in classes including linear algebra, multivariable calculus, and engineering mathematics.

Master Of Data Science

Master Of Science

B. Arts Sc. Honours