Abstract
The growing presence of computational technologies across both professional and everyday environments motivated the two parts of my thesis: a technical capstone focused on an AI-driven document automation system and an STS research paper examining TikTok’s role in shaping fashion microtrends. For my capstone, I helped to develop a machine learning supported document processing tool during my consulting internship at PricewaterhouseCoopers (PwC). In contrast, my STS research investigates how TikTok’s For You Page (FYP) and microtrend cycles influence fashion self-expression among young women by amplifying certain styles. The connection between these projects is that both examine how designed technical systems shape human behavior. In one case by structuring professional workflows, and in the other, by directing attention and influencing cultural participation. Bringing these projects together highlights that engineering is not only about building systems that function effectively, but also about understanding how those systems shape the ways that people think, act, and make decisions.
The technical portion of my thesis produced a machine learning supported document processing and automation tool designed to streamline consulting workflows within PwC. Developed during my Technology and Analytics internship in the Financial Services Risk and Regulation practice, the system uses a modular, agent-based architecture to automate repetitive tasks such as data extraction, document formatting, and report generation. It uses AI agents coordinated through a Model Context Protocol server to provide a standardized and auditable interface for tool invocation and data exchange. This design allows the system to process both structured and unstructured data while maintaining consistency with internal templates and regulatory requirements. Using this tool improved workflow efficiency and reproducibility by reducing manual processing time and minimizing variability in outputs.
In my STS research, I examined how TikTok’s FYP and algorithmically driven microtrend cycles influence fashion self-expression among young women. Drawing on Actor-Network Theory (ANT), I analyzed how both human actors (influencers, users, and fast fashion brands) and non-human actors (the FYP, engagement metrics, and production systems) interact to produce and amplify fashion trends on TikTok. My research argues that microtrends on TikTok do not emerge organically, but are instead constructed through a self-reinforcing system of algorithmic visibility, influencer promotion, user participation, and rapid fast fashion production. Within this system, repeated exposure to specific aesthetics makes them appear widely adopted and culturally urgent, encouraging users to participate while narrowing the range of styles that feel socially acceptable. This limits opportunities for individual self-expression and contributes to a cycle of constant trend turnover and overconsumption. While this research does not empirically confirm a direct impact on self-esteem, it demonstrates how these dynamics can weaken the relationship between clothing and personal identity.
From an STS perspective, ANT emphasizes that technologies do not operate independently, but instead function within networks of human and non-human actors that shape outcomes. In my technical capstone, the document automation system was designed to improve efficiency and an explicit STS perspective was not incorporated into its design. However, applying ANT reveals that such systems do more than streamline workflows, they also influence how consultants interact with information, structure their work, and make decisions. This is important because STS research shows that technical systems are not neutral, they actively shape human behavior. Although this perspective was not considered in the design of the document automation tool, it highlights the need to account for these influences in professional environments. This same dynamic is reflected in my STS research, which shows that TikTok’s platform design and visibility structures do not simply reflect user preferences, but actively shape cultural participation and perceptions of identity. Taken together, these projects show that engineering decisions go beyond technical performance to shape autonomy, behavior, and social norms, highlighting the importance of STS perspectives in designing technologies responsibly.