Samira Kaur
I work with messy data, uncover patterns, and turn them into decisions people can actually use. I'm currently studying Data Science at Wilfrid Laurier University with a minor in Physics.
About Me
I'm a Data Science student at Wilfrid Laurier University, minoring in Physics and based in Waterloo, Ontario.
My work focuses on machine learning, financial risk, and making technical results understandable to the people using them. I'm especially interested in building analyses that are not only accurate, but practical enough to support real decisions.
Physics has shaped the way I approach data problems by teaching me to look past surface-level complexity, find the underlying structure, and question whether the explanation actually fits the evidence.
Outside the classroom, I'm involved in student leadership and technical education. Through Case Hacks, I helped organize a Scotiabank-sponsored hackathon, and I currently serve as VP of Internal Affairs at the Laurier Analytics Society.
Skills
Experience
Director of Education
I help shape the educational side of Case Hacks, from workshop planning to case development. Most recently, I supported a Scotiabank-sponsored hackathon and helped students use tools such as OpenAI Codex and Google Stitch to move quickly from an idea to a working prototype.
Vice President of Internal Affairs
I coordinate internal operations for a 15-person team, keeping projects organized and helping members work effectively across events, workshops, and other initiatives.
Content Director
I created beginner-friendly Python and SQL content for student workshops and helped organize a case competition hosted at Google. The role taught me how to make technical material feel approachable without stripping away the parts that matter.
Notes
Thoughts, write-ups, and things I'm learning. Coming soon.
Notes coming soon.
Projects
Things I've built, mostly in Python; credit risk, NLP, segmentation, and dashboards. Click a title to view the source on GitHub.
Cost-sensitive credit risk pipeline built on 5,700+ borrower records. Compared four classification models, tuned decision thresholds using expected loss, and applied SHAP to translate predictions into lending policy recommendations.
Built a retrieval-augmented Q&A system that searches internal documents and returns citation-grounded answers. Evaluated retrieval, answer quality, and citation accuracy across a structured test set.
Segmented 1,000+ customer records using K-Means, PCA, and silhouette analysis. Identified six behavioural groups that could support more targeted customer and marketing strategies.
Built a market-monitoring workflow that classified 5,200+ records into stress levels using SQL logic. Connected the output to an interactive Excel dashboard for tracking volatility, market spikes, and key indicators.