Resume

Education

Duke University | 2022-2024

Enhanced and expanded functionality of the Lenovo Workstation Diagnostic app (Android & iOS), ensuring compatibility with Android 13 and the latest iOS version.
Implemented support for Windows desktop app Lenovo Performance Tuner, incorporating Windows 11 compatibility and introducing new tuning features and performance analysis capabilities.
Contributed to full-stack development of the Lenovo Workstation website, creating and managing APIs and web-based tools for both internal and public web projects.

Conducted experiments using network diagnosis tools to measure the performance of commercial 5G network and simulate the scheduling algorithm of the base station.
Data analysis/visualization using python scripts, using libraries such as pandas and seaborn. Collected and processed real world coordinates into GeoTile data. Generated reports on Wireless 5G signal strengths based on geo locations and network load.

Designed python scripts applying machine learning and computer graphics and using Tensorflow framework to implement an automated masking program. Processing video data using automated annotation tool to mask any sensitive content in the raw data and enhance the video quality. Collected and processed video data can be found in Ego4D project.
Process, label, and upload data to UMN supercomputer, utilized automated scheduling algorithm to facilitate process efficiency.
Lead recruitment and video processing for the Minneapolis/St. Paul area. Host virtual meeting with participants and distribute/recycle experiment tools such as video capture glasses and equipment calibration. Coordinate in-person experiments with motion capture and video equipment.

Developed Accent ID system for the lab’s projects, the revised model after training later is integrated in to the NSFC program for voice recognition.
Collected various of accent audio data with geo-specific infromation. Hands on experience using machine learning and NLP to achieve automated system. Applied multiple modules to create a classification script that could distinguish different dialects using their characteristic wave matrix.