Experience

USC Viterbi School of Engineering

• United States
• Higher Education
• 700 & Above Employee

• Course Assistant

  • Jan 2022 - Present

  EE538(Computing Principles) Course Teaching Assistant • Instruct weekly 1.5-hour in-person discussions to review data structures and algorithms • Collaborate with other TAs to prepare teaching materials with Google Slide • Hold weekly 2-hour office hours to help over 100 enrolled students debug C++ code with Bazel • Design and grade assignments and final project with unit testing in GoogleTest to test students’ C++ knowledge EE538(Computing Principles) Course Teaching Assistant • Instruct weekly 1.5-hour in-person discussions to review data structures and algorithms • Collaborate with other TAs to prepare teaching materials with Google Slide • Hold weekly 2-hour office hours to help over 100 enrolled students debug C++ code with Bazel • Design and grade assignments and final project with unit testing in GoogleTest to test students’ C++ knowledge



USC Ming Hsieh Institute - Department of Electrical and Computer Engineering

• United States
• Higher Education
• 1 - 100 Employee

• Research Student

  • Dec 2021 - Present

  Direct Research Assistant • Build a Spike-Element-Wise residual block and the Threshold-Dependent batch normalization in the Residual Spiking Neural Network, achieving an accuracy of 89.91% so far • Build and train an ANN model with the same residual block to be used as pre-trained model in SNN • Fine-tuned P2M-DeTrack model (a custom faster-RCNN model), achieving mAP value (IoU=0.5-0.95) of 33.2% using MMDetection framework and COCO-style protocol on the BDD100K multiobject tracking validation dataset • Optimized Parametric Leaky Integrate-and-Fire model with updated-by-layer threshold, achieving a 97.22% test accuracy on DVS128 Gesture dataset



Vincho, Inc

• Los Angeles, California, United States

• Machine Learning Intern

  • Jun 2022 - Aug 2022

  • Finetuned and trained Real-ESRGAN_x4plus model to be used for artistic image restoration and upscaling • Trained Neural Style Transfer model by VGG19 with artistic dataset to output images with general Vincent-style filter • Finetuned and trained Real-ESRGAN_x4plus model to be used for artistic image restoration and upscaling • Trained Neural Style Transfer model by VGG19 with artistic dataset to output images with general Vincent-style filter



HiRain Technologies

• China
• Motor Vehicle Manufacturing
• 1 - 100 Employee

• Test Engineering Intern

  • Sep 2020 - Nov 2020

  Assisted and performed ECU testing for three Geely automotive models by programming in CANoe environment to test and adjust software and hardware accordingly Assisted and performed ECU testing for three Geely automotive models by programming in CANoe environment to test and adjust software and hardware accordingly