Hello! I’m Jonathan Cheng

M.S. Computer Science Student @Colorado University at Boulders.

• YouTube
• LinkedIn

About me

Projects

WordPress block patterns offer a powerful way to create predefined block

Research Software Engineer @Academia Sinica

Table of Contents

• Simplify content creation and ensure design consistency.
• Streamlines the design process and saves time.
• Pre-arranged collections of blocks.

Intro

Life at Academia Sinica, the largest research institution in Taiwan, as a Research Software Engineer is both challenging and rewarding.

My primary responsibility is conducting research in information security, specifically using patient data to predict diabetes while ensuring patient privacy. This involves generating synthetic data that retains the main features of the original data without disclosing sensitive information. Thus, producing high-quality synthetic data is a crucial task. In our research, we employ techniques such as autoencoders, variational autoencoders, and GANs (Generative Adversarial Networks) to help generate the synthetic data.

The mechanism behind these deep learning methods might seem complex, but here’s a simple analogy: imagine downloading a zip file. When you zip a file, it compresses the data to save space, similar to how an autoencoder’s encoder compresses input data into a smaller representation known as the bottleneck. This bottleneck retains the essential features of the original data, like how a zip file maintains most of the original content’s key characteristics. When you unzip the file, it returns to a format close to the original, akin to how the decoder reconstructs the data from the bottleneck. The extent to which features are preserved depends on the bottleneck’s size: a higher dimension in the bottleneck retains more features, much like a higher resolution in a zip file.

By utilizing synthetic data, we successfully addressed both information security and diabetes prediction challenges. I thoroughly enjoy constructing the research framework, coding, and fine-tuning hyperparameters.

The concept of these

Beyond Work

A walk with the other lab mates in the afternoon always refreshes my mind and gives me new ideas.

Is research life getting frustrating? Unhealthy Friday to the rescue!

There you go!

Led the implementation of survey question logic, text structure, and styles for online surveys using HTML, CSS, and Python

• Details: I was responsible for creating and organizing the logical flow of survey questions to ensure they were clear, concise, and easy to follow. This involved designing the structure and layout of the survey using HTML and CSS to make it visually appealing and user-friendly. Additionally, I used Python for backend logic to handle survey responses and data processing.
• Impact: These efforts enabled us to gather actionable consumer insights, which contributed to generating over $100,000 in revenue annually by improving the quality and effectiveness of the surveys.

Architected a new analytic model for brand loyalty measurement with client’s need fulfillment

• Details: I designed and implemented a comprehensive analytical model to measure brand loyalty. This model took into account various metrics and client-specific needs to provide accurate and meaningful insights into customer loyalty and satisfaction.
• Impact: By reducing analysis time by 30%, the model not only increased efficiency but also provided timely and relevant insights to clients. My contributions were recognized with the “Worker of the Month for Inclusivity” award in Q2 2022, highlighting my ability to work effectively within diverse teams and meet client needs.

Developed RESTful APIs to serve data to a JavaScript front-end based on user inputs

• Details: I created robust and efficient RESTful APIs that interacted with the backend database to fetch and serve data dynamically based on user inputs. This allowed for seamless communication between the server and the JavaScript front end, enhancing the overall user experience.
• Impact: The APIs successfully handled requests from over 3,000 users, demonstrating their reliability and scalability. This development was crucial for enabling real-time data access and interaction, significantly improving the functionality of the web application.

Optimized SQL scripts to automate manual reporting processes

Impact: This automation saved the business approximately 25 hours each month, freeing up valuable time for team members to focus on more strategic initiatives.

The efficiency gains also contributed to more timely and accurate reporting.

Details: I examined the existing manual reporting processes and identified opportunities for automation using SQL scripts. By optimizing and automating these scripts, I streamlined data extraction

, transformation, and reporting tasks.

Beyond Work

Lunch breaks are always a great time for some gossip, which helps recharge my energy!

The trip to Yilan was wonderful!

After collaborating with the Hong Kong branch for two years, I finally got to meet my colleagues there during a trip to Hong Kong. It felt just like meeting online friends in real life!

At the last supper, My manager said, “Three of you will betray me,” to his team members.

Three of us were set to leave Nielsen within the same month, and our manager joked that this was the last supper. It was a beautiful, warm night, filled with memories, laughter, and tears. The heartfelt goodbyes made this evening even more special.