Integrated and implemented a data fetching pipeline into a proprietary machine learning model evaluation platform, helping automating the inline training and evaluation process.

Developed a machine-learning-based anomaly detection model, trained on production line images.

Optimized the quantization flow of the algorithm, reducing 50% of computational resource usage.

Optimized memory usage from over 20 MB to 55kB by re-implementing the algorithm in a streaming-based approach, making it possible to be executed on the chip.

Holistic analysis on filter flow and quantization factor optimization.

Skills: Computer vision algorithm, C/C++, python, Linux, quantization analysis

Development of AR/VR optical intrinsic and extrinsic parameter calibration tool.

Development of holistic image quality evaluation tools for AR/VR devices, i.e., MTF, Color Aberration, Field of View, Distortion, Virtual Image Distance.

Designed logging framework for optical inspection on AR/VR production lines.

Responsible of initial design of AR and VR head-mounted display

Designed test plans and test procedure in the initial development stage of each product

Familiar with lens manufacturing process and quality control, data analysis of product’s yield rate

Defined optical specifications for head-mounted displays

Developed optical test projects with customers and provide customers with optical solutions

Novel AR and VR patent research

Products involved: Windows Mixed Reality series, HP Reverb, HP Reverb G2, Google Glass II, Third-Eye AR