Biography

I grew up in Pingtung, Taiwan. Due to the limited resources in this rural area, I had an intense desire to access more opportunities and face bigger challenges from a young age. I was introduced to computers and the internet early on. At 14, I built a website using Microsoft FrontPage, offering MP3 downloads at a time when copyright issues weren't yet a concern, and we were still using 56k dial-up modems to connect to the Internet.

As my teenage years unfolded, my life was fueled by rock'n'roll. Playing in bands as a main vocalist and guitarist, I immersed myself in the music of Oasis, Guns N' Roses, Radiohead, Nirvana, and Beatles. This passion guided me through my formative years and eventually (kind of weirdly) led me to National Cheng Kung University in 2003, where I majored in Engineering Science. While my studies covered a broad range of topics, it was computer science that truly captivated me. My work on a neural network project using Self-Organizing Maps for image segmentation solidified my love for software development.

Recognizing Taiwan's strength in hardware but its weakness in software, I decided to pursue further education abroad in the UK. I earned a master's degree in Artificial Intelligence from the University of Southampton, followed by a PhD in Robotic Vision at Imperial College London. My research focused on Visual Simultaneous Localization and Mapping (VSLAM), Deep Learning, and high-performance programming languages like C/C++ and CUDA. This period marked my deep dive into solving real-world problems using advanced technologies.

After completing my PhD, I joined University College London as a Postdoctoral Research Associate, contributing to the EU-funded CASCADE project aimed at advancing autonomous robotic surgery. It was here that I realized the profound gap between academic research and real-world application. This realization ignited my journey into entrepreneurship.

My first startup, Umbo Computer Vision, sought to revolutionize video security through real-time computer vision. This experience illuminated a critical gap in handling unstructured data—an area with insufficient tooling and often requiring expensive AI and data teams.

Determined to address this challenge, I founded Instill AI to provide more efficient solutions for unstructured data processing. As Generative AI, large language models (LLMs), vision language models (VLMs), and large multimodal models (LMMs) emerged in 2022, it became clear that while these technologies made significant strides, the software tooling needed to keep pace.

I am on it.

Publications

Check out my Google Scholar profile for more details.