cv

M.S. in Computer Science

University of California, Los Angeles

B.S. in Computer Science and Information Engineering

National Taiwan University

• Overall GPA - 4.14/4.3
• Ranking - 13/140

Software Engineer Intern

Microsoft

• Implementing a full-stack application to automate the procedure of stress testing, which enables engineers to stress multiple servers simultaneously and monitor the status through rack managers.
• Optimized Kusto and MSSQL database procedure to expedite data fetching efficiency by 17% and the amount of data from 10M to 100M+.
• Reconstructed the Azure data dashboard to help engineers analyze the Azure hardware event log.

IT Intern

Funpodium

• Implemented a decision-tree model to determine the order and type of advertising based on user's behavior, which boosted the user return rate by 16%.
• Collected application logs from company's apps and constructed a dashboard to visualize the user flows of apps.

Section Chief, Wireless Services, Network Administration Division

NTU CSIE Network Administration and System Administration Team

• Led a 4-man team to maintain and troubleshoot routers and wireless access points of department CSIE to ensure stable and secure wireless connection.
• Managed the ELK stack system in CSIE department to store the event logs from other devices.
• Conducted a large-scale documentation update and a cleanup of existed setting of Radius and Zone Director.

Speech Processing and Machine Learning Lab

NTU, Professor Hung-Yi Lee

• Collaborating with Professor Hao Tang, University of Edinburgh.
• Member of 2022 Eighth Frederick Jelinek Memorial Summer Workshop in Johns Hopkins University (join virtually).
• Researched different compression methods on self-supervised speech models.
• Proposed a new model architecture that achieve similar performance to state-of-the-art model while reducing the training time by 30% and VRAM consumption by 50%.

Machine Intelligence and Understanding Lab

NTU, Professor Yun-Nung (Vivian) Chen

• Researched natural language processing (NLP), with focus on dialogue state tracking.
• Leveraged controllable counterfactuals data augmentation to improve accuracy of baseline model by 7% in accuracy.
• Investigated the possibility to conduct membership inference attack on NLP translation model.

Skull Fracture Detection

• Integrated the architecture of YOLO and CenterNet to detect the area of skull fracture, which yielded 64% accuracy in F1 score.

Sports Team System

• Built a full-stack system to manage the badminton team.
• Improved the efficiency of the rental of badminton courts.

Don’t Touch the White Tile

• Built a Java music game from the game of the same name.
• Structured the game with objected-oriented design such as open-closed principle to achieve great extensibility.

Travel Recommendation Chat Bot

• Designed a chat bot that recommended user where to travel based on user's input text.
• Used Microsoft Bot Framework as backbone and utilized LUIS to provide human-like interaction.
• Was sponsored by Microsoft to complete the project after CareerHack hackathon.

• Dean's List Award