Staff Algorithm Engineer (ML)
LinkedIn | Google Scholar | Github
Email: twcmchang [at] gmail [dot] com
Hi, I'm Chun-Min (Jimmy), originally from Taiwan and currently working as an algorithm engineer in the U.S.
My expertise lies in model quantization, pruning, and mixed-precision inference, focusing on accelerating large language models (LLMs) and deep learning systems. Before moving to the U.S., I collaborated with Sheng-Wei Chen at Academia Sinica, applying deep learning technologies to advance healthcare in Taiwan. My research has been published in Nature Communications, npj Digital Medicine, and IEEE International Joint Conference on Neural Networks (IJCNN).
Beyond my professional work, I am passionate about surfing. My favorite spots include The Hook and Steamer Lane in Santa Cruz, Linda Mar and Rockaway in Pacifica, and Ocean Beach in San Francisco. I share my thoughts and life lessons on Threads.
Staff Algorithm Engineer (ML) at Ambarella Corporation, Santa Clara, U.S.A., 2020/06 - Now
- Build machine learning compilers for large language models (LLM) and general deep learning models.
- Design user-friendly programming language and tools for our SoC platforms.
- Improve model efficiency and retain accuracy on SoCs by optimizing bit width usages and operations.
Machine Learning Engineer at Cardinal Blue Software (PicCollage), Taipei, Taiwan, 2019/01 - 2019/07
Design and deploy lightweight deep learning models on iOS and Android platforms for photo style transfer, photo super-resolution, low-light photo enhancement, and object instance segmentation.
Research Assistant at Academia Sinica (alternative military service), Taipei, Taiwan, 2015/10 - 2018/12
Apply cutting-edge machine learning techniques to interdisciplinary studies, and publish research works in Nature Digital Medicine, Nature Communications, Computer methods and programs in biomedicine.
Chin-Chi Kuo, Chun-Min Chang, Kuan-Ting Liu, Wei-Kai Lin, Hsiu-Yin Chiang, Chih-Wei Chung, Meng-Ru Ho, Pei-Ran Sun, Rong-Lin Yang, and Kuan-Ta Chen
npj Digital Medicine 2019 / Paper
We developed a deep learning approach for automatically determining the estimated glomerular filtration rate (eGFR) and chronic kidney disease (CKD) status over 4,505 kidney ultrasound images. Overall CKD status classification accuracy of our model was 85.6% —higher than that of experienced nephrologists (60.3%–80.1%).
Sipher Wu*, Chun-Min Chang*, Guan-Shuo Mai, Dustin R Rubenstein, Chen-Ming Yang, Yu-Ting Huang, Hsu-Hong Lin, Li-Cheng Shih, Sheng-Wei Chen, and Sheng-Feng Shen
Nature Communications(*equal contribution) 2019 / Paper
Using over 20,000 images with GPS locality information belonging to nearly 2,000 moth species, our approach learned deep representations that accurately predict each species’ mean elevation based on colour and shape, as well as help identify the underlying mechanisms generating this biogeographic pattern.
Chun-Min Chang, Chia-Ching Lin, and Kuan-Ta Chen
IEEE Joint Conference on Neural Networks (IJCNN) 2019 / Paper
We proposed a novel training algorithm for convolutional neural networks to achieve dynamic inference at run time and enhance model robustness against adversarial attacks without any extra computational cost or memory overhead.
Chun-Min Chang, Cheng-Hsin Hsu, Chih-Fan Hsu, and Kuan-Ta Chen
ACM International Conference on Multimedia (ACM MM) 2016 / Paper
We propose the very first non-intrusive measurement methodology for quantifying the timing and positioning accuracy of commodity Virtual Reality (VR) systems. Our methodology considers the VR system under test as a black-box and can work with any VR applications without code instrument or system modification.
Chun-Min Chang and Kuan-Ting Liu
Collaboration with Applied Science Institute, Academis Sinica, 2018
We leveraged signal processing and machine learning to transform in-the-wild sensor data from the wearable device into intelligent health experiences. Training over 1,200 records from 600 fasting participants, our non-invasive approach showed over 93.5% of test samples in zone A of the Clarke consensus error grid.
Chun-Min Chang, and Bin Li
Collaboration with Biomedical Engineering, University of Wisconsin-Madison, 2019
We design a unsupervised model for differentiating histology tissue matrices. The learning algorithm is guided by superpixel refinement and deep feature clustering, and beats other unsupervised methods in metrics of accuracy and inference time.
