Jingyi (Duncan) Cheng

Nanjing,Jiangsu

Summary

AI/ML Engineer with experience in industrial data processing, machine learning evaluation, computer vision, and AI application development. Proficient in Python, Scikit-learn, TensorFlow/Keras, OpenCV, and SQL, with hands-on experience building data pipelines and supporting machine learning workflows. Passionate about applying AI to predictive analytics, anomaly detection, and intelligent manufacturing.

Overview

years of professional experience

Work History

AI Tech Intern

Epoch Information Development Co., Ltd.

Nanjing, Jiangsu

07.2023 - 09.2023

Deployed and evaluated open-source AI models for a digital human project, conducting inference testing, case-based performance analysis, and technical reporting to support model selection.
Investigated and benchmarked open-source AI models including So-VITS-SVC, SadTalker, Wav2Lip, Stable Diffusion, and rembg for speech-driven animation and image processing applications.

IT Consultant Co-op

QA Consultants (ALTEN Group)

Toronto, Canada

01.2022 - 05.2023

Weld Defect Simulation and Detection System

Developed a simulation-based data generation pipeline using OpenSCAD and FEMM, generating over 10,000 labeled defect samples to support downstream machine learning development.
Co-developed a standardized machine learning evaluation pipeline using confusion matrices, ROC/PR curves, AUC metrics, and threshold analysis to support model evaluation by an external AI team.

HD Map Data Validation Tool

Built a Python-based data processing pipeline using NetworkX and OpenCV to validate HD map topology through multi-source map comparison and identify disconnected road segments.
Evaluated positional accuracy for 57,574 traffic signs across 274 categories and developed a Dash + Leaflet dashboard for visualization and quality analysis.

Education

Master of Electrical and Computer Engineering - Project-based

University of Western Ontario

London, Canada

10-2025

Bachelor of Computing - Hons., Specialization in Artificial Intelligence

Queen's University

Kingston, Canada

06-2024

Skills

Programming: Python, SQL, TypeScript, JavaScript, C
Machine Learning: Scikit-learn, TensorFlow/Keras, OpenCV, PyTorch
Data Analysis: NumPy, Pandas, Jupyter Notebook

Frameworks: React, Nextjs, Django, Nodejs
Databases & Tools: MySQL, SQLite, Git, Docker, Google Maps API, Prisma

Selected AI Projects

Transit AI Planner for Urban Cities | Queen's QMIND AI Research Team

Machine Learning Developer | Oct 2022 – Apr 2023

Developed an AI-assisted transit planning system in a five-member team using OpenStreetMap API to collect and process geographic and points-of-interest (POI) data across six cities.
Applied Scikit-learn K-Means clustering, 5th-order polynomial regression, and a custom Gaussian scoring algorithm to model transportation demand and optimize transit routes.
Built an interactive visualization platform using React and Leaflet, achieving a service score of 69.75 compared with 44.15 for the Toronto baseline under a 1 km walking coverage metric.

Wildfire Evacuation Visualization Platform | Western University MoTech Research Group

Full-Stack & AI Developer | Jan 2025 – Sep 2025

Developed a full-stack wildfire evacuation platform using React, Next.js, TypeScript, Google Maps API, Prisma ORM, and SQLite to visualize wildfire zones, shelters, and evacuation routes.
Designed and implemented 10+ RESTful APIs, integrating JWT authentication and a constrained Dijkstra algorithm for risk-aware evacuation routing.
Integrated a Dify AI workflow to provide structured wildfire information retrieval based on user-selected locations and time.
Explored Q-learning and Dueling DQN under faculty supervision to investigate reinforcement learning approaches for dynamic evacuation route optimization.

Timeline