Hsiao-Ting Huang
AI Security · Embedded Systems · Trustworthy ML
Portfolio
Software Engineer @ MediaTek · M.S. ECE (University of Washington)
Interests: Trustworthy ML · Computer Vision · NLP · Embedded Systems
About Me
I am a software engineer with a strong background in machine learning, computer vision, and time-series modeling, as well as hands-on experience in embedded systems and modem technologies. My career goal is to develop secure, reliable, and robust AI-enabled systems that tightly integrate sensing, decision-making, and actuation, particularly in safety-critical and resource-constrained environments.
Selected Projects & Results
AI Video Captioning
Developed a dense video captioning solution for 1-hour recordings, bridging visual and language data for indoor & outdoor scenarios. Enhanced the PDVC model for event detection and captioning, enabling users to monitor daily events and identify critical situations in long-duration video footage.
Computer Vision Video Captioning
TRIPDECODER
Proposed a novel approach to recover metro commuter routes not captured by AFC systems. Decoupled travel time inference and route preference estimation, achieving higher accuracy and efficiency than prior works. Validated with large-scale datasets from Singapore and Taipei.
ACM/IMS TDS Data Science
Air Pollution Real-Time Prediction
Designed a spatial-temporal model (LSTM+CNN+SVM) for AQI forecasting, reducing MAE by 27% compared to state-of-the-art. Ranked top 7% in KDD Cup 2018. Built web and mobile applications to display real-time air quality data for Beijing and London.
KDD Cup Top 7% Time Series Machine Learning
Ainimal
Cofounded an AI-powered dating app leveraging NLP to match users based on communication style. Built sentiment analysis models using Jieba for Chinese tokenization, TF-IDF for word frequency, and LinearSVC for feature extraction.
Cofounder NLP Machine Learning
Skills
ML/Data: Computer Vision, Time-Series Forecasting, NLP
Systems: Embedded Systems, Modem Technology, Algorithm Optimization
Programming: Python, C/C++, PyTorch, TensorFlow
Looking Ahead
- Robust AI perception under sensor perturbations
- Secure & efficient ML on embedded/edge devices
- Reliability evaluation in safety-critical systems
Contact
- Email: tristina.hth@gmail.com
- LinkedIn: linkedin.com/in/hsiao-ting-huang
- GitHub: github.com/tristinahuang