Hao-Hsuan Chang, Ph.D.

Staff Research Engineer | Samsung Research America

📧 haohsuanchang2918@gmail.com

📄 Curriculum Vitae 🎓 Google Scholar 💻 GitHub 💼 LinkedIn

Staff Research Engineer at Samsung Research America specializing in AI solutions for wireless sensing and communication systems. With over 8 years of experience, I develop low-complexity AI applications optimized for commercial embedded devices, with expertise in time-series prediction, reinforcement learning, recurrent neural networks, and LLM finetuning. I have a strong track record of driving projects from concept to deployment, resulting in high-impact publications, patents, and commercial products.

Education

Ph.D. Virginia Tech, Electrical and Computer Engineering | Aug 2017–Aug 2021

M.S. National Taiwan University, Communication Engineering | Sep 2013–Jun 2015

B.S. National Taiwan University, Electrical Engineering | Sep 2008–Jun 2013

Selected Projects

AI-Powered Sleep Monitoring System

Designed and implemented AI models for sleep stage prediction through fusion of mmWave and UWB radar sensing, incorporating ResNets, transformers, and multi-modal learning. Developed efficient wireless signal processing pipelines in C++ for human localization, motion tracking, and sleep monitoring on embedded devices. Commercialization expected in 2027.

Activity Recognition with mmWave Sensing

Created AI models for activity recognition and adult/kid classification using mmWave sensing technology. Developed classification algorithms for human motion detection and activity pattern recognition with high accuracy and low computational complexity.

WiFi-Based Multi-Person Respiration Monitoring

Developed CSI amplitude and phase compensation techniques to enable real-time multi-person respiration rate tracking with WiFi signals. Created advanced preprocessing algorithms for anomaly removal and signal compensation, achieving accurate detection of multiple distinct respiration rates from single antenna pairs.

Intelligent Human Presence Detection

Designed WiFi-based human presence detection algorithms for indoor environments using two-stage detection systems combining macro-motion sensing with micro-movement and breath signal analysis. Implemented robust occupancy detection with RF interference filtering and motion classification.

Dynamic Spectrum Access with Distributive Deep Reinforcement Learning

Designed distributive multi-agent deep reinforcement learning approaches for spectrum sharing and load balancing optimization. Developed lightweight recurrent neural network architectures for time-series prediction of multi-user spectrum access behavior, resulting in 7 publications.

Technical Skills

Certification: AWS Certified Machine Learning Engineer, AWS Certified Cloud Practitioner

Programming: PyTorch, TensorFlow, ONNX, SQL, Python, Matlab, C++

Deep Learning: LLM finetuning, Transformers, Reinforcement Learning, RNNs, CNNs, XGBoost

Selected Patents

Commercial impact through AI-powered wireless sensing and health monitoring solutions

Multi-antenna WiFi based breathing rate estimation - Advanced CSI impairment compensation and multi-dimensional analysis for accurate multi-person breathing rate estimation using WiFi signals.
Wifi csi preprocessing for sensing applications - Intelligent preprocessing framework that corrects device impairments and filters interference to enhance WiFi sensing accuracy and reliability.
Wi-fi based human presence and motion detection - Two-stage detection system combining macro-motion sensing with micro-movement and breath signal analysis for robust occupancy detection.
Respiration rate detection using wi-fi - Anomaly removal and signal compensation techniques enabling accurate detection of multiple distinct respiration rates from single antenna WiFi CSI data.
Presence detection and recognition with Wi-Fi - Advanced CSI processing with anomaly removal and compensation algorithms for reliable multi-person presence detection and respiration monitoring.