Joshua Lee

Joshua Lee

Machine Learning Engineer & Racket Sports Enthusiast

Recent grad from University of California, San Diego with a Bachelor of Science in Data Science. I like building things that make sense of messy data — whether that's detecting faults on a production line or making medical imaging more interpretable. Outside of work I'm usually on a pickleball court or checking out new food places with my friends.

Projects

Edge-Deployable Keyword Spotter

Edge-deployable 10-keyword spotter that compresses a DS-CNN to under 100 KB through INT8 quantization (PTQ and QAT), with sub-millisecond inference through ONNX Runtime.

PyTorch ONNX Runtime Quantization Edge AI Keyword Spotting

FusionTrack: Multi-Sensor UAV Tracking with EKF and Multi-Object Tracking

2D UAV tracker that fuses radar and camera through an Extended Kalman Filter, with a Global Nearest-Neighbor tracker for crossing-target scenarios.

Python Extended Kalman Filter Sensor Fusion Multi-Object Tracking

AeroTrack: Aerial Multi-Object Detection, Tracking, and MLOps Pipeline

End-to-end aerial perception pipeline that detects and tracks people and vehicles in drone footage with YOLOv8 and ByteTrack, served behind FastAPI.

YOLOv8 ByteTrack Computer Vision FastAPI MLflow

Semiconductor Yield Fault Detection

End-to-end fault detection pipeline on SECOM manufacturing data, from imbalanced-class modeling through MLflow-tracked experiments to a Streamlit demo.

PyTorch scikit-learn MLflow FastAPI Streamlit

Explainable Medical AI for Pulmonary Edema Detection

Explainable deep learning system that combines CNNs and a medical LLM to detect pulmonary edema from chest X-rays and paired radiology reports.

PyTorch Computer Vision NLP Grad-CAM

ISR Drone Deployment Trade Study

Scenario-based operations analysis of ISR drone fleet deployment under cost, coverage, and persistence constraints.

Python Simulation Optimization Operations Research

WeatherWear: AI-Powered Outfit Recommendations

AI agent that classifies real-time weather with a Random Forest model and recommends full outfits through a rule-based policy.

Python scikit-learn FastAPI Random Forest

Bikewatching

Interactive geospatial visualization of Bluebikes traffic flows across Boston and Cambridge over the course of a day.

JavaScript D3.js MapLibre GL Geospatial

Meridian Hospital Analytics: Healthcare Operations Analytics Case Study

Healthcare operations dashboard surfacing patient, provider, appointment, procedure, and billing activity in a hospital-style analytics view.

Python SQL Dashboard Data Visualization

Experience

Skyworks Solutions

Digital Analytics & Content Operations Co-op

Jan 2026 – Present

Built a privacy-first analytics pipeline that generates 0-100 Content Performance Score (CPS) for Skyworks web pages using cookie-lite engagement metrics (scroll depth, dwell time, bounce, exit rate).

Industrial Technology Research Institute (ITRI)

Data Analyst Intern

Jul 2025 – Aug 2025

Designed an NLP pipeline for large-scale text data, including data cleaning, feature extraction, and structured signal generation.

The Floc App

Data Science Intern

Jul 2025 – Aug 2025

Assisted in building iOS-based social matchmaking app for Gen Z users.

Maxzone Auto Parts Corporation

Data Engineering Intern

Jun 2024 – Sep 2024

Automated reports and supported data operations across departments.

Code Ninjas

Coding Instructor

Jun 2021 – Jul 2021

Taught coding fundamentals and led a stop-motion animation workshop for students.

Education

University of Southern California

Master of Science, Applied Data Science

May 2026 – Dec 2027

University of California, San Diego

Bachelor of Science, Data Science

Sep 2022 – Mar 2026