1. Developed a Python-based AI platform for real-time applications, mapping and conversion between common large models in ONNX and hardware-specific IR.

2. Developed a C++-based compiler to align with auto-driving chip requirements, enabling efficient code generation, simulation, and quantization.

Worked for Huawei Technologies Co., Quality, Business Process IT Mgmt Dept.

• Designed an appearance inspection system, and implemented the automatic detection of dozens of connector appearance defects with deep learning algorithms.

• Developed an object detection system to detect type of equipment or meter reading based on images collected by cameras.

• Dealt with the missing data of the personal record part of the 2017 Hawaii population data using GUIDE, EM algorithm, multiple imputation method (MICE and Amelia package), regression imputation method and cross-validation.

• Predicted the average value of the variable INTP with inverse probability weighting method and direct prediction method.

• Improved the method of prediction using LASSO and cross-validation for logistic regression with inverse probability weighting method predicting the mean value of INTP, making it closer to government estimated value and thereby more reliable .