Summary
Ting-yao Hsu is an AI research scientist with a decade of hands-on experience in machine learning, multimodal vision-and-language systems, large language models, and human-AI interaction. Currently at TSMC after completing a PhD at Penn State, he has blended academic rigor with industry experience through internships at Amazon, NEC Labs, and Megagon Labs working on LLM agents, data-to-text, and summarization. His doctoral and research-assistant work produced novel Seq2Seq/transformer post-editing for visual storytelling, a first human-edited visual story dataset of ~15k stories, and graph-convolution approaches to model sequential image semantics. Comfortable shipping prototypes to mobile and web, he also has roots in systems and storage research from earlier roles at Academia Sinica and National Tsing Hua University, reflecting a rare cross-disciplinary fluency between low-level systems and high-level multimodal AI. Colleagues describe him as a pragmatic researcher who moves models into interactive products while keeping human preferences front and center.
10 years of coding experience
9 years of employment as a software developer
Doctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at Penn State University
Bachelor's degree, Compter Science, 3.5/4.0, Bachelor's degree, Compter Science, 3.5/4.0 at 國立清華大學
Chinese, English, 西班牙文