• Built trading assistant app using React Native, Redux, and TypeScript with an experimental approach of A/B test.
• Developed API server for stock transaction information with Node.js, Express, GraphQL.
• Improved query using GraphQL and Apollo to cut down response time by 10%.
• With 72% accuracy, Predicted the volumes of trading based on how many times is the stock cited on Reddit, using beautiful soup.

Software Development Engineer Intern

• Constructed a program to calculate the complexity of brain images (Alzheimer's) in MATLAB.
• Preprocessed the fMRI images and analyzed time-series images, to extract the features of images according to the complexity.
• Statistically verified the validity of the data (ANOVA) and shortened the time by 12%, utilizing Python.

Data Scientist Intern

• Researched computer vision models for the dental imaging software ('EzDent-I') and got FDA approval.
• Advanced pipelines for Computer Vision models on hardware accelerators for optimal throughput, latency, and power
consumption; performance increased by 30%
• Converted C++ Computer Vision APIs to Python to make it easier for users to use and built unit test code with 80% coverage.
• Contracted training time by 20%, leveraging Cloud-Based Data Infrastructure (Min.io, Docker, Kubeflow, and Grafana Labs).

AI Researcher, Vatech

Dec 202
• Through LSTM analysis, optimized the public bicycle sharing system (time-series data) and reduced the cost to arrange the bicycles by 10% (dropping RMSE to 1.742).

Research Student

With python, automate implementing tools (Calibre, starRC, FineSim), predict the result by regression analysis, and shorten the experiment time by 20%.

Electrical Engineering Intern

Find out detailed information about byungsoo choi,  in Los Angeles, CA

byungsoo choi: in Los Angeles, CA

byungsoo choi

Ziggurat Technologies

USC Neuroimaging and Informatics Institute

latency, and power

Korea University

Samsung Electronics

Korea University

University of Southern California