• Developed AWS Neptune Data Plane SDK using Smithy IDL and AWS SDK codegen libraries from scratch. Lead working across teams including AWS SDK teams and Neptune Product team. The developed SDK supported 45 APIs in 14 programming languages. The Data Plane SDK will enable both internal and external developers and users to better interact and integrate with AWS Neptune DB.• Developed AWS Neptune Data Plane SDK Customer Tests using cucumber BDD test tool for features and parameterized tests in different programming languages. Worked on the API documentation using AWS doc package.• Demoed the SDK in JS, Java using AWS Neptune, S3 and EC2 to mimic external user behaviors. Gave final presentation to the Neptune team. Provided all the related docs.

• Participated in supporting and expanding the Library’s geospatial enterprise initiatives and assisted with technical issues, create and maintain scripts, write technical documentation • Responsible for notebooks testing, debugging, transforming and modifying(re-devlopement) for UCSD Geospatial projects and DSC170 course material on different platforms• Developed a Flask framework and using ajax for collecting the questionaries on LimeSurvey for DSC170 Admin Server

Automated UV Drone Disinfection, advised by Prof. Tara Javidi• Modified the front-end design of the project website, refitted the parrot Drone hardware and added UVC disinfection function• Implemented math-based off-line energy saving and obstacle avoidance coverage path planning via python• Detected the drone in the video and images using Yolo V4 and used Jupyter notebook as PyTorch model inference environment. Implemented drone-obstacle distance approximation using OpenCV• Engaged in the imitation learning part of the project including indoor field data collection, algorithm simulation, and comparison.

• Learned and participated in developing a web application including user interface and business functions with Bootstrap and Spring MVC• Responsible for database maintenance and implemented create, read, update, delete support for code entities using MySQL

• Optimized the distribution of the asset packaging and cleaned more than 250,000 customer information data in Python• Completed Lorenz curve and stats graph, established customer level classification model, and used python pandas to distribute the customer assets and define fitness level calculation rule• Log transform to set the fitness level of running samples to normal distribution, using Z test to set min fitness level into an optimal interval, fairly distribute the asset packages to the managers