William Glazer-Cavanagh Email & Phone Number

Vancouver, BC, CA

Vancouver, British Columbia, Canada

• Developping state of the art models to address unsolved problems in the field of antibody discovery
• Developed a novel semi-supervised learning pipeline, significantly reducing labeling costs and improving model performance.
• Engineered efficient data structures for image processing, leading to substantial speedups in training and inference times.
• Designed reusable abstractions for common ML tasks, streamlining workflows for the team and fostering collaboration.
• Conducted in-depth error analysis and provided actionable insights to stakeholders, driving model improvement and strategic decision-making.
• Developed and deployed robust diagnostics tools and visualizations to monitor model performance, identify anomalies, and facilitate iterative improvements.

Montreal, Quebec, Canada

• Automated high throughput screening of potential molecules using state of the art models
• Designed systems which outperformed pre-existing models in 80% of the images in out of distribution (OOD) settings as measured by human preference

Montreal, Quebec, Canada

• Optimization and evaluation of call center processes across Canada
• Increased profits by implementing optimal call center strategies in production. Evaluated diverse methods with our discrete event simulator.
• Enabled early detection of problems in production by developing dashboards and ETL pipelines to monitor our models.
• Detected defects and encouraged re-use of tree-based model features by quantifying their impact on predictions

Montreal, Quebec, Canada

• Natural language processing (NLP) research project to automatically explain stock price history from news articles
• Improved company’s machine learning pipeline by integrating state of the art NLP algorithms
• Maximized reproducibility and reduced bottlenecks by refactoring the algorithm evaluation pipeline to a workflow orchestration tool
• Identified avenues of research by assessing the state of the art in knowledge injection with knowledge graphs for NLP algorithms. Discussed interesting findings with principal researcher to establish a research agenda

Montreal, Quebec, Canada

• Development of Merck Canada’s internal ETL (extract, transform, load) processes for Business Intelligence
• Produced first insights on GARDASIL-9 HPV vaccine’s canadian market by spearheading a data science opportunity forecasting project. This project was used to power a canada-wide marketing and awareness campaign against.
• Initiated DevOps best practices for deploying our processes, reducing delays to push code to production by 66%
• Automated Merck Canada’s manual ETL workloads and reduced delays by 2 days on average by automatically ingesting data from external vendors.

Montreal, Quebec, Canada

• Contributed to the IVADO/MILA online Deep Learning class. This class covered the most popular types of deep learning models. Hired after volunteering at the IVADO/MILA Deep Learning School

• Taught a group of students the fundamentals of Data Science throughout a semester

Montreal, Canada Area

• Analysis of GitHub Machine Learning repositories to extract insights on best practices from open source models and libraries from a software engineering standpoint.
• Started the project by designing the case studies, then collected data by web scraping GitHub. The project was picked up by post-docs in the MCIS lab after my departure and led to an accepted EMSE paper.

Master'S Degree, Machine Learning

Bachelor'S Degree, Software Engineering - Artificial Intelligence And Data Science

Pre-University College Diploma, Pure And Applied Sciences

High School