Public Playlist

Experience

Jeffrey Peng · 3 songs, 12 min 21 sec

#Title

Loblaw Digital · Toronto, ON

Led development of a Python-based LLM system for invoice anomaly detection to reduce $300M in annual maintenance spend by 10%, integrating SQL and Looker dashboards to visualize model insights for business teams.
Built a batch-processing system in Python (JSONL + LiteLLM) with prompt caching to operate within API rate limits, cutting request volume from 600,000+ to 600 and enabling cost-efficient, scalable processing on enterprise data.
Engineered and selected features for LLM invoice classification, improving model accuracy from 62% to 85%.

September - December 2025

Miovision · Kitchener, ON

Built and tuned logistic regression, decision tree, and random forest ML models to predict traffic collisions using scikit-learn, pandas, numpy, and matplotlib, helping city planners identify high-risk intersections.
Created dashboards and reports in Looker, leveraging SQL to join, filter, and transform raw traffic data into meaningful visual insights for stakeholders.
Developed a Python package for traffic flow modeling, signal optimization, and traffic model validation.

May - August 2024

University of Waterloo - W Store · Waterloo, ON

Optimized W Store product filtering by creating and populating new index tables with SQL and BASIC, cutting query time from 2.3 seconds to <0.01 seconds for 40,000+ students.
Owned the end-to-end design and development of a new web application that allowed professors to submit an exam print form using PHP, MySQL, AJAX, jQuery, and HTML/CSS.
Modernized Docker dependencies for Magento upgrades, improving system stability and deployment efficiency.

January - April 2023