I am Chung-Yi Chen (陳忠義), a first-year Ph.D. student in Computer Science at National Yang Ming Chiao Tung University, advised by Prof. Yi-Ping You. I am also a security compiler engineer at ICEshell Co., Ltd., leading the development of ICEshell’s compiler-based software protection solution.

• M.S. or Ph.D. in Computer Science and Engineering, Institute of Computer Science and Engineering, National Yang Ming Chiao Tung University (2022/09 - Present)
  • I went directly to the Ph.D. program in August 2024 as my master’s thesis research had satisfying outcomes. If I am lucky enough, I will receive the Ph.D. or, otherwise, the M.S. degree.
• B.S. in Computer Science and Engineering, Department of Computer Science and Engineering, National Sun Yat-sen University (2018/09 - 2022/06)
  • I graduated with the third highest grade in the department!

I am interested in the intersection of compilers and machine learning. My ongoing research is an ML-friendly program representation for LLVM IR. A domestic workshop, CTHPC 2024, accepted the primary results of this research. The proposed representation is named rand4vec, which can give you a vector representing a program. rand4vec is compute-efficient, explainable, flexible, and training-free. I believe rand4vec is a promising representation that can be integrated with LLVM to develop practical applications of machine learning methods. I plan to submit this research to ACM TACO, and hope you can find my paper there soon!

It might be weird that a compiler guy has a solid machine learning background, but well, this is who I am. I took several undergraduate and graduate courses on machine learning. During my first year of university, I led a team of my friends to participate independently (i.e., without any assistance from the university) in the 2019 AI CUP Competition on Biomedical Paper Analysis - Biomedical Relation Extraction Track hosted by the Ministry of Education and achieved the Excellence Award (fourth place). I also wrote an article about the mathematical foundation of reinforcement learning, which you can find on my Medium page.

I also have experience with AI accelerators. A series of articles on my Medium describe how I implemented a systolic array-based AI accelerator on FPGA and accelerated an image classification model with hardware software co-optimization. This solution is the fastest on the course Accelerator Architecture for Machine Learning 2023, which achieved a 1.8x speedup compared to the second fastest or a 10x speedup compared to the unaccelerated baseline. When we met at the CTHPC 2024 venue, Prof. Tsung Tai Yeh, the lecturer of the course mentioned above, told me that he was really surprised by the performance of my solution and even asked his master’s students to investigate my design.

If you seek collaboration for your research, please contact me via [email protected].

ICEshell is a security startup based in Taipei that aims to provide high-quality software protection solutions for mobile app developers. Our investors include the well-known APP development company Sparkful (formerly Fourdesire) and the world-leading security company DEVCORE.

I have been responsible for the iOS APP protection solution since February 2024. After several months of effort, this compiler-based solution has been delivered to a TPEx-listed (over-the-counter) company for proof of concept. I attended several cybersecurity exhibitions to introduce our solutions, including CYBERSEC 2024, Taiwan, and CyberDSA 2024, Malaysia. The next exhibition I will attend is the Taipei Game Show 2025. Besides, I represented ICEshell in taking group photos with Ms. Hsiao, Vice President of the Republic of China, at CYBERSEC 2024.

• To many people’s surprise, I like singing and am good at it (not boasting). You may hear me playing guitar and singing in the laboratory on some lonely nights : )
• I started playing video games when I was seven years old. Diablo-like and RTS games are my favorites.