Abstract
My technical project and STS research both examine a commonality within engineering
problems, that technologies often promise access while quietly placing new work on the people
they are supposed to help. In my technical project, this problem appears in the University of
Virginia’s course scheduling process, where students navigate between SIS, Lou’s List, The
Course Forum, Reddit, advisors guidance, sites, spreadsheets, and notes on what classes are
required, necessary, and fits within schedule. All of this just to build a workable semester
schedule. In my STS research, a similar issue appears in digital accessibility, where blind and
low vision users are often told that a website, document, or AI tool is “accessible,” even when
they must interpret confusing outputs, results, verify uncertain information, or create work
arounds. Although these projects address different domains, both are honestly concerned with
how engineering decisions distribute effort, confidence, and risk. STS matters to engineering
because it shows that a system’s success cannot be measured only by whether it technically
functions. Engineers must also ask who the system serves, who it burdens, what assumptions
are made, and whether it allows users to participate with greater independence.
The technical portion of my thesis has produced Cavalier Calendar, an AI powered interactive
course scheduler designed to reduce the cognitive burden of planning a students semester. The
project responds to recurring student frustrations with SIS, including session expirations, the
need to know exact department codes, laggy navigation, fragmented course information, missed
prerequisites, and undetected conflicts. Rather than replacing SIS or handling direct enrollment,
Cavalier Calendar acts as an intelligent planning layer that helps students conceptualize their
schedule before final registration. Its distinguishing feature is a conversational interface that
allows students to ask natural language questions such as “Find me a CS class related to app
development that fits within my schedule”. The system is designed to use public SIS-style
course data, validate prerequisites, detect time conflicts, surface enrollment and waitlist
information, and support saved schedule comparison. The project also emphasizes accessibility
though a keyboard friendly interface, clear restrictions of interface elements, plain language
explanation of academic jargon, and user testing with UVA students. The potential significance
of this project is that it reframes scheduling from a code based search task into a guided
decision making process. If successful, Cavalier Calendar could help first year students, transfer
students, first generation students, students with disabilities, and students exploring unfamiliar
departments make confident academic choices with less friction.
In my STS research, I examined how accessibility for blind and low vision users is often treated
as a compliance outcome rather than a lived experience. My paper argues that accessibility
should be judged by reliability, user control, and burden distribution, not by formal conformance
or technical capability alone. I focused on the gap between standards based accessibility and
the actual experiences of users navigating websites, PDFs, screen readers, overlays, and AI
based accessibility tools. New AI systems can describe images, extract text, interpret scenes,
and provide conversational assistance, but these tools also introduce problems of hallucination,
trust, verification, affordability, and dependence on platforms or connectivity. My research
synthesizes accessibility studies, AI accessibility literature, and STS concepts to show that tools
can appear innovative while still shifting labor onto its users. The significance of this work is it
provides a stronger framework for evaluating accessibility, not a simple, “Does this tool provide
extra help?” but “Can users depend on the help without excessive labor, uncertainty, or loss of
control?”
Considering the technical, organizational, and cultural elements of both projects together
clarifies what ethical engineering requires. From an STS perspective, neither a scheduling
assistant nor an accessibility tool is only a technical artifact, each is part of a large network of
users, institutions, data sources, policies, interfaces, economic constraints, and assumptions
about what users should already know or be able to do. Cavalier Calendar depends not only on
code, APIs, and AI models, but also on UVA’s scheduling structures, student advising practices,
FERPA concerns, course availability, and the unequal familiarity students bring to academic
planning. AI accessibility tools similarly depend not only on whether blind and low vision users
are meaningfully included in design. Together, these projects taught me that responsible
engineering should reduce hidden labor rather than disguise it. A well-designed system should
not merely perform a task; it should make participation more dependable, understandable, and
fair for the people who rely on it.
