Abstract
My thesis portfolio explores two separate sociotechnical problems, rather than one shared one.
My technical project focuses on the problem of making orbital data understandable for
non-experts, while my STS research examines the problem of how fairness is socially
constructed in AI hiring systems. Even though these projects are unrelated in subject matter,
they both show how technical systems are never just technical. They are shaped by the people
who design them, the institutions around them, and the way users experience them. In my
technical work, the larger issue is that space data is widely available but still inaccessible to
students and the public because existing tools are too technical or not educational enough. In
my STS work, the larger issue is that AI hiring systems are often presented as neutral even
though the meaning of “fairness” is decided by employers, vendors, laws, and auditors. Both
problems matter because they show how technology can either widen or reduce gaps in
understanding, access, and opportunity depending on how it is built and governed.
My technical project investigates how to lower the barrier to understanding Earth’s increasingly
crowded orbital environment through an educational web platform called OSCAR (Orbital
Surveillance & Collision Assessment Radar). The specific problem is that existing satellite and
near-Earth object trackers either prioritize professional accuracy with super complex interfaces
or they simplify the experience so much that it loses educational value. To address this, my
team and I designed a platform that combines real TLE and NASA NeoWs (both satellite data),
a physics propagation engine (to predict satellite movement), time controls, and observer based
visualization into one accessible software. We analyzed evidence through usability testing,
performance metrics, physics validation, and task completion rates that were directly tied to our
user stories. Our most important finding is that combining accurate live data with simple
visualization tools can make orbital mechanics and satellite information much easier for
students and non experts to understand without having to be overly technical or knowledgeable.
The project showed that educational design choices like cartoonish views, guided views,
informative interactables and metadata are just as important as the code and physics models
used in helping users build an idea of how satellites move.
My STS research investigates how fairness in AI hiring systems is defined, stabilized, and often
made difficult to challenge and adjust. The specific problem is not simply whether the algorithms
are biased, but how organizations decide what fairness even means in the first place. Using
Actor Network Theory and SCOT< I studied how fairness emerges through networks of
employers, engineers, vendors, datasets, audits, and legal systems. I analyzed scholarly
papers, policy documents and real world cases such as Amazon’s hiring tool, New York City’s
Local Law 144, and the Workday lawsuit. My most important finding is that unfairness persists
because fairness is reduced into small metrics, audits, or compliance steps that become the
objective truth. Once these definitions are integrated into the systems, they become hard for
candidates to question, even when applicants still experience bias. This research showed that
fairness in AI hiring is not a fixed technical property, but a sociotechnical outcome that is shaped
by power, governance, and different actor experiences.
Overall, I believe both projects successfully contributed to their own broader sociotechnical
problem frames, even though they approached very different issues. My technical project
contributed by making complex orbital systems more understandable and educationally useful
for all audiences. My STS project contributes by showing why fairness problems in AI hiring
cannot be solved through technical fixes alone. At the same time, both projects have limitations.
OSCAR still depends on device performance, internet access, and the predictive limits of SGP4
over long time windows. My STS work is limited by its focus on existing cases and policies
rather than direct interviews with candidates or hiring practitioners. Future technical researchers
should continue improving accessibility, scalability, and educational testing for orbital
visualization systems. Future STS researchers should focus more directly on candidate
experiences and accountability that goes beyond just audits and compliance. Together, these
projects strengthened my understanding that technologies only become meaningful when we
also study how people use them, interpret them, and are affected by them.
I would like to sincerely thank my capstone teammates Elliot Hong, Thomas Welch, and Kyle
Vitayanuvatti for their collaboration on OSCAR, as well as my technical and STS advisors for
their guidance throughout this process - Professor Wiley and Professor Sherriff. I am especially
grateful to my paper reviewers who challenged my ideas and helped me improve my STS paper
as well.
