CV
Contact Information
| Name | Patrick Deng |
| Professional Title | Data Engineer & MLOps |
| keqi.deng@outlook.com | |
| Phone | 647.514.4114 |
| Location | Toronto, Ontario |
| Website | https://kdeng4.github.io/ |
Professional Summary
Data Engineer building production pipelines and applied ML systems. M.Sc. Computer Science, Georgia Tech.
Experience
-
2024 - Mississauga ON
Data Engineer (MLOps)
Walmart Canada
Own the end-to-end Ads CRM data warehouse on GCP BigQuery, processing 70 M+ records daily to power customer targeting, segmentation, and marketing ROI.
- Architected and own the Ads CRM data warehouse in GCP BigQuery, orchestrating ETL pipelines via Airflow that process 70 M+ records daily and serve as the single source of truth for customer targeting.
- Designed a customer segmentation framework using clustering and dimensionality-reduction techniques, enabling marketing to launch precision campaigns that measurably improved ad ROI.
- Built an identity-resolution pipeline linking web and in-store transactions, increasing identifiable customer events by ~2 % and strengthening cross-channel attribution.
- Developed a GenAI agentic tool with LangGraph and LLM APIs (GPT, Gemini) that automates data-quality checks and ad-hoc analysis, reducing manual investigation time.
- Conducted statistical analyses (regression, time-series, A/B testing) on large-scale customer datasets, delivering actionable insights to product and marketing stakeholders.
- Partnered with data-science and analytics teams to optimize back-end data sources, improving accuracy and simplicity of downstream reporting workflows.
-
2021 - 2024 Mississauga ON
Data Engineer
Dane Creek Capital Corp.
Built the cloud data platform from the ground up for a multi-brand B2C/B2B company, delivering BI dashboards, predictive models, and cost-efficient infrastructure.
- Designed and implemented a cloud-based data warehouse on GCP with Airflow, integrating CRM, ERP, and analytics sources into scalable ETL pipelines that improved reporting accuracy by 25 %.
- Deployed Metabase as the company-wide BI tool, replacing costly legacy software and reducing data-tooling licensing costs by ~40 %.
- Built predictive models for customer behaviour analysis (churn prediction, product recommendations) that improved retention rates and informed product-offering decisions.
- Designed a loyalty-program data framework, structuring customer engagement metrics and redemption tracking to support strategic marketing initiatives.
- Optimized financial forecasting workflows with Python, SQL, and analytical tools, reducing manual effort and improving operational efficiency across the finance team.
- Championed adoption of GCP and Azure cloud services, establishing cost-efficient, scalable infrastructure for the growing data team.
Education
-
2022 - 2025 Atlanta GA, USA
Master of Science
Georgia Institute of Technology
Computer Science - Interactive Intelligence
- Artificial Intelligence
- Knowledge Based AI
- Natural Language Processing
- Database System Implementation
- Game AI
- Software Development Process
- Machine Learning for Trading
- Intro to Information Security
- Human Computer Interaction
-
2018 - 2020 Calgary AB, Canada
-
2014 - 2018 Calgary AB, Canada
Publications
-
2026 Exploring Transitions of Graduates From an Online Master's in Computer Science Program to Doctoral Programs
SIGCSE 2026
Presented at SIGCSE 2026. Analyzes how Georgia Tech’s OMSCS program prepares graduates for STEM PhD programs through enrollment data, surveys, and interviews. Demonstrates that affordable, asynchronous online programs can effectively support non-traditional students pursuing doctoral education.
Skills
Languages
Interests
Projects
-
Raven's Progressive Matrices (RPM) Agent
Designed an AI agent to solve Raven’s Progressive Matrices, a human intelligence test, using knowledge-based AI techniques. Achieved 80% accuracy on advanced visual analogy problems.
- Implemented an AI agent to process visual analogy problems in 2x2 and 3x3 formats
- Reflected on human cognition by modeling AI to achieve human-like reasoning
-
American Sign Language Recognition with Hidden Markov Models
Designed and implemented an AI agent leveraging Hidden Markov Models (HMMs) to recognize American Sign Language (ASL) gestures from video input.
- Trained HMMs on a dataset of ASL signs, achieving accurate recognition of gestures with high confidence levels
-
Job Application Timeline Tracker (Android App)
Developed a full-featured Android app to track job application timelines, enabling users to organize, edit, and monitor applications from start to finish.
- Designed a user-friendly interface for managing job applications
- Utilized IntelliJ IDEA for development and Android SQLite for data storage
- Followed a comprehensive software development process, including design, test planning, and documentation