Abstract
This portfolio presents two interconnected research projects that together examine enterprise technology from complementary angles: one technical, one social. The capstone project addresses the reliability of enterprise software systems, specifically the design and execution of a structured testing framework for a large-scale SAP implementation. The STS research paper examines the environmental sustainability of the AI and cloud infrastructure that powers systems like SAP, asking how the expansion of AI and data centers has increased electricity demand and carbon emissions, and to what extent current sustainability strategies can reduce those impacts. On the surface, one project is about making enterprise systems work reliably before they go live, and the other is about the hidden environmental costs of running them at scale. The connection runs deeper than that: both projects are ultimately about the gap between what technology promises and what it delivers, whether that is a promised go-live date free of defects, or a promised commitment to sustainability free of accountability. Together, they reflect the reality that enterprise technology must be evaluated not only by whether it functions correctly, but by what it costs the world to run. The capstone project addresses a problem that affects nearly every large organization: what happens when a complex enterprise system like SAP fails because errors were not caught before deployment. Large-scale ERP implementations carry significant risk if system errors or misconfigurations go undetected prior to go-live. To address this risk, the project developed a structured software testing framework during an internship in SAP technical consulting. The work involved designing detailed test scenarios that translated business requirements into verifiable technical cases, covering SAP order-to-cash workflows and SAP and Salesforce integration. Testing included functional, integration, end-to-end, and edge-case scenarios, with particular focus on integration points where data crossed system boundaries, since those areas presented the highest risk of failure. The process required collaborating with business stakeholders, and different technical teams to align requirements with test coverage and to communicate results in terms accessible to non-technical audiences. The outcomes included early detection of data mapping errors, pricing calculation problems, concurrency issues, and API performance failures that would have caused significant disruption if left unresolved until after deployment. Beyond the immediate project, the work produced a reusable testing library and reinforced practices in verification, validation, and enterprise systems integration that are broadly applicable to future implementations.
The STS research paper examines the environmental burden of the AI and cloud infrastructure that underlies modern enterprise systems. The expansion of AI and data centers has measurably increased both electricity demand and carbon emissions, driven by the compounding demands of computing workloads, cooling infrastructure, hardware manufacturing, and the pace of global digital growth. While a range of technical sustainability strategies have been developed and validated in research, including task scheduling optimization, workload consolidation, carbon-aware dispatch, and machine learning-based energy management, their adoption remains uneven and insufficient to offset the rate of AI-driven growth. The paper applies the Social Construction of Technology framework to explain why. SCOT reveals that the gap between available solutions and their implementation is not primarily a technical failure; it reflects the contested nature of what sustainable AI means across stakeholder groups with divergent interests. Cloud platform providers, enterprise customers, researchers, regulators, and local communities each attach different meanings to sustainability, and the groups with the greatest power to resolve that contest have structural incentives to favor narrow definitions that minimize compliance burden. The paper concludes that meaningful progress requires not only better engineering, but governance frameworks that hold dominant actors accountable to comprehensive sustainability standards. Working on both projects simultaneously produced insights that neither project would have generated on its own. The capstone work involved building systems that enterprise clients trust to perform reliably under real-world conditions, and part of that work meant confronting the infrastructure requirements that reliability demands redundant servers, continuous uptime, high-volume data processing. That day-to-day technical reality made the STS research feel less abstract. The energy consumption figures in the STS paper stopped being statistics and started representing the actual systems being tested, configured, and deployed. At the same time, the STS research sharpened the capstone work by raising questions about accountability that pure technical practice tends to skip over. In consulting and software testing, success is measured by defect counts and go-live readiness. The STS framework pushed toward a broader definition of success, one that includes asking who bears the cost of running these systems and whether the institutions governing that cost are adequate. The most valuable outcome of working on both projects is a clearer sense of what responsible engineering requires. It is not enough to build systems that work. The field also requires asking what it takes to run them, who decides what those costs are worth, and whether current answers to those questions are good enough.
