Abstract
The technical capstone project, developed in collaboration with Rayyan Alam and Suzu Paudel, focused on creating a lightweight GPU scheduling platform that makes shared computing resources easier to access and manage. The system was designed to help students and researchers run GPU-based workloads through a simple web interface without requiring advanced technical setup or command-line experience.
The project combines frontend and backend development, containerized job execution, user authentication, and resource management into a unified platform. Users can submit and monitor jobs through the web application, while the scheduler handles execution order, runtime limits, and system safety features automatically. Overall, the project emphasizes accessibility, usability, and reliable GPU resource sharing in small academic or research environments.
The STS research paper examines the growing use of artificial intelligence in government decision-making and the ethical and social challenges that come with it. As public agencies increasingly rely on data-driven systems to guide decisions in areas such as housing, transportation, and public safety, the paper explores how these technologies influence the way communities’ needs are identified and prioritized.
Using the Social Construction of Technology (SCOT) framework, the paper argues that AI systems are not fully neutral or objective because they are shaped by the values, assumptions, and biases of the people and institutions that create them. The research discusses how algorithmic systems can unintentionally reinforce existing inequalities, even when designed to improve efficiency and planning. At the same time, the paper acknowledges that AI can provide meaningful benefits when paired with human oversight and responsible implementation. Overall, the research highlights the importance of considering the social impact of technology in policy and decision-making.
These two projects focus on very different topics. The STS paper studies how artificial intelligence and algorithms influence government decision-making, especially in situations that can affect large communities. The technical project, in contrast, is a practical tool designed to help students and researchers share GPU resources more easily. The projects were completed independently and were not intentionally designed to connect with one another. However, both projects share a common idea: making access to important resources fair and accessible. The STS paper looks at how algorithmic systems can unintentionally disadvantage certain groups when biases exist in the data or design process. The GPU scheduler addresses a smaller-scale version of a similar challenge by making shared computing resources easier to use for people without advanced technical experience.
The projects also connect through their focus on design choices and usability. Features such as the web interface, simplified setup process, and user-friendly controls in the GPU scheduler were created to reduce barriers for less technical users. In this way, both projects highlight how technology design can shape who is able to access and benefit from a system.
