• Developing NNTrainer, an open source light-weight on-device training framework (https://github.com/nnstreamer/nntrainer).

• Develop and revamp C programming assignments that apply in-depth knowledge of system programming and fix any potential bugs.• Enhance students' knowledge of system, including process control, synchronization, signals, memory management, and shell programming.

• Led perception team to develop real time lane detection system for autonomous vehicle using Tensorflow and ROS.• Performed project of object detection and tracking for autonomous vehicles using Pytorch.• Designed a motion planning algorithm for autonomous vehicle decision-making with implementation.• Participated in implementing models of human drivers in CARLA to make an autonomous vehicle simulator more realistic. • Advised by Prof. Alireza Talebpour

• Conducted research on quantization methods and various neural networks for efficient inference on object detection.• Implemented various quantization algorithms on the device, measure the performance, and analyze the result.

• Contributed to Intel OpenVINO development, an open-source toolkit for optimizing and deploying AI inference as a member of the IOTG ODT OpenVINO Runtime Team.• Enhanced the accuracy and performance of deep learning models of Intel OpenVINO.• Formulated parallelization of a compilation of inference engine configuration in OpenVINO using OpenCL.• Managed implementation, performance optimization, and testing of deep learning primitive kernels for Intel GPU architectures.• Responsible for analyzing Intel's pre-trained deep learning model architectures and demos such as image segmentation to develop tools for analyzing deep learning models.

• Engineered radar control system development to control multiple radars simultaneously.• Conducted specification-based tests and unit tests of the vessel traffic service operating system.• Designed an algorithm to track unregistered vessels from radar data through data analysis and visualization and integrated into a vessel traffic service operating system.• Managed software maintenance to convert the analog signal radar data into digital data in vessel traffic service radar image extraction system.