I am a 4th-year student at the University of California, San Diego, pursuing a Master of Science in Computer Science after having completed my Bachelor of Science in Computer Science in June 2025. I specialize in driving user and results-oriented innovation by identifying complex challenges, constructing strategic roadmaps toward nuanced solutions, and leveraging my product expertise to strategize, analyze, market, and develop impactful products that shape the world. My research interests focus on employing adversarial/deceptive planning and strategic obfuscation within reinforcement learning agents.

I am actively pursuing New Grad opportunities in Product Management, Software Engineering, and AI/ML, and am eager to showcase my skills and gain insights into industry-standard practices. Ultimately, my goal is to transition into a Product Management role, aligning with my deep-seated interest in advancing data-driven technology and leveraging my innate abilities in external communication and leadership.

Selected Experience

Professional Experience

As a Software Engineer Intern at OnePay in New York City during Summer 2025, I focused on advancing cloud-based AI infrastructure and customer interaction workflows. I redesigned the RAG v.2 bot’s Post Contact Analysis pipeline by integrating OpenAI’s batch API with fp-ts (functional programming in Typescript), which reduced operational costs by 50% and improved fault tolerance across over 10 million voice and chat interactions. I also pitched and led the integration of GenAI workflows to address resolution delays, aligning Engineering, Design, and Product Ops teams to deploy a scalable solution with AWS Lambdas that fostered a steady 4% month-over-month increase in customer satisfaction. In parallel, I enhanced observability in the RAG bot pipeline by analyzing DynamoDB metadata logs and tracing LLM API outputs, reducing debug time by approximately 64% and cutting failure rates by 18% as part of ongoing QA and fault-tolerance initiatives. Additionally, I drove Scrum Agile practices by facilitating sprint planning, leading backlog grooming, and prioritizing tickets for the customer interactions team, enabling the delivery of eight new features in just nine weeks while maintaining a sprint completion rate above 90%.

As a Software Engineer Intern at GymBuddy during Fall 2023, I led a cross-functioning team consisting of other software engineers, UX designers, and GTMs, as an unofficial Engineering Manager in collaboration with the UC Berkeley UX Club to enhance existing web applications while working in a remote setting. I strategically managed project workflow with JIRA and ensured feature quality through continuous integration and deployment (CI/CD) pipelines with GitHub Actions. Notably, I also prototyped an Advertiser Dashboard web application using the MERN stack, featuring a frontend with React.js, while deploying the backend infrastructure with Google Firebase, Express.js, and Node.js to ensure a seamless user interface and facilitate data queries for advertising partners. Additionally, I conducted A/B testing on user interactions with Google Analytics, validating a 28% increase in active user traffic, and applied these insights to guide feature improvements.

As a part-time Software Engineer at Endeverus in San Mateo, CA during Fall 2022, I specialized in developing baseline RESTful APIs and backend systems using the Django framework in Python, focusing on integrating core functionalities and ensuring seamless communication between user event handlers and data queries for web applications. Furthermore, I integrated EC2 Servers with cloud-based data through the AWS platform within automated systems, optimizing data flow and enhancing the efficiency of client applications. In addition, I pitched into product-centric decisions such as designing the user interface, employing quality assurance with CI/CD pipelines, as well as managing the user experience of client components.

Research

I was previously an Undergraduate Computer Graphics Machine Learning Researcher as a part of the ERSP program at UCSD Computer Science and Engineering Department under the advisory of Prof. Ravi Ramamoorthi and Prof. Tzu-Mao Li from the Center for Visual Computing and Differentiable Rendering Research Group. I led a team of four in evaluating Slang, a new differentiable shading language, through efficiency and performance analysis, in collaboration with Nvidia.

Research proposal: Image View Synthesis and Differentiable Rendering in Slang
Research poster: Evaluating Slang for View Synthesis

Another research project I worked on was Assessing the limitations of QAGNN (Question Answering using Language Models and Knowledge Graphs) in which I proposed the implementation of a Natural Language Inference model to verify that the model’s predicted answer would follow the logic of a question, as well as addressing ambiguous priority of question-answer groups by re-ranking answers with an MRR-based (Mean Reciprocal Rate) Neural Network. The papers and codebase can be found in this repository: QA-GNN Analysis Study.

Projects

Throughout my time at the University of California, San Diego, I have conducted a large series of team projects. Of such projects, the one I am most proud of is Codelog, a user-centric developer journal aimed at illuminating and sharing the intricate journey of passion projects. In this initiative, I led a software engineering team of ten student engineers, emphasizing agile methodologies to manage sprints using JIRA, implementing CI/CD pipelines for code maintenance, and conducting full-stack development with HTML, CSS, and JavaScript. This project was my initial introduction to software engineering and product management at a localized level, providing me with fundamental baseline knowledge that I have now carried into my current and future roles.

In a more individual passion project, I crafted a KNN Forecast Machine Learning TradingView Indicator and Algorithm. Here, I created a K-Nearest Neighbors regression model to forecast future price points, yielding a remarkable 2461.42% return on TSLA and 767.49% return on NVDA. Additionally, I enhanced a third-party script using Python, Skopt, and Selenium with a new feature to refine parameter selection, contributing to open-source trading optimization: TradingView Machine Learning Scripts

More About Me

On the side, I am a Realtor based in the San Francisco Bay Area, driven by my commitment to thrive in our family business. I am an avid volunteer when it comes to giving back to my previous communities.

Some experiences include serving as a Web Administrator for my Virtual Tutoring SF initiative during the COVID-19 Pandemic, where I scaled my nonprofit’s ability to provide free tutoring resources and helped oversee operations that resulted in over 11,000 classes taught, 1,280 students supported, and 400 tutors engaged.

At UC San Diego, I held the position of Social Chair for the Taiwanese American Student Association, where I organized events that strengthened cultural awareness and community engagement on campus, which amassed event sizes that ranged from local events to regional gatherings.

I have also been an Assistant Meet Director with the California Interscholastic Federation San Francisco Section, where I worked as a certified Track and Field Official to ensure smooth meet operations and fair competition.

Additionally, I have contributed as an Assistant Varsity Football Coach at Lowell High School, focusing on linebackers, secondary, and defensive quality control, where I emphasized discipline, technical skills, and mentorship to help athletes succeed both on and off the field.