About Me

Hi there, welcome to my web Portfolio

I am GuangCheng, a Junior Undergrad at University of Michigan -Ann arbor, studying Computer Science. My interest and passion lies in the fields of Machine learning, and Software Engineering

I strongly believe that new technologies and software can be embraced by society as a whole and offer individuals more opportunities and convenience. With this in mind, I aim to immerse myself in the industry, engage with these advancements, and contribute to their ongoing development.

Experience

Work Experience

Machine Learning Engineer @ Hong Kong Applied Science and Technology Research Institute (Astri) 6/2024-present

During my summer intership at Astri's IoT Sensing and AI Technologies department, I helped Intelligence machine vision(IMV) team deploy object detection model(YOLOv7). My contribution included implementing parallel data processing for generating training set, switching to GPU for model training, data augmentation, generating and feeding randomly augmented data into the model every epoch to improve robustness and adding deformable data samples to detect deformable object

HTML5 Bootstrap Template by colorlib.com HTML5 Bootstrap Template by colorlib.com HTML5 Bootstrap Template by colorlib.com HTML5 Bootstrap Template by colorlib.com

Math Teaching Assistant at University of Pittsburgh 08/2022-12/2022

As a math Teaching Assistant, I offered recitation(tutoring lesson) once a week for the college Algebra class and held office hours, answering questions from walk-in students at the Math Assistant Center, supporting over 50 student in different math classes

HTML5 Bootstrap Template by colorlib.com

Projects

Deep learning model for image classification task 01/2024-05/2024

Built a deep learning model and trained a classifier that distinguishes two different places of interest.

Identified the optimal model architecture,adjusted hyperparameters, applied regularization techniques and experimented with various data augmentation methods for superior performance

Implemented transfer learning. The pretrained model are trained from a much larger dataset that included 8 other places of interest. Pretrained model provided a good starting point for training and the model converge faster

HTML5 Bootstrap Template by colorlib.com HTML5 Bootstrap Template by colorlib.com

no Pretraining

66.23%

transfer learning

89.25%
HTML5 Bootstrap Template by colorlib.com HTML5 Bootstrap Template by colorlib.com

Search Engine 01/2024-05/2024

Code snippets

Designed a MapReduce pipeline to efficicently process and crawl web content.

Started mutiple server to further analyze and process the data outputed from the pipeline

Applied parallel data processing to fetch information from different server and find top 10 search result

Developed web interface with REST API using Flask and successfully deployed applications to AWS

HTML5 Bootstrap Template by colorlib.com HTML5 Bootstrap Template by colorlib.com HTML5 Bootstrap Template by colorlib.com HTML5 Bootstrap Template by colorlib.com
Get in Touch

Contact

HTML5 Bootstrap Template by colorlib.com

gavinggc@umich.edu

HTML5 Bootstrap Template by colorlib.com

+1 412 277 8217

© Copyright All rights reserved. Made with by Colorlib Distributed by ThemeWagon Demo Images: Unsplash.com