ABOUT
ME

Arizona State University (Barrett Honors) | Graduating Spring 2026

This is my GCSP portfolio . The NAE Grand Challenges Scholars Program is a national program where engineering students work across five competencies (Talent, Multidisciplinary, Entrepreneurship, Multicultural, and Social Consciousness) to prepare themselves for the biggest problems out there. In this portfolio, you will find my research on AI safety and ambiguity in LLMs (Talent), interdisciplinary analyses through FSE 150 and STS 330, my entrepreneurial work with Venture Devils, multicultural perspectives from HON 272 and ASB 378, and service-learning through the EPICS program with Mayo Clinic which is all connected by my theme of Security.

My theme is Security, chosen because my academic work consistently returns to questions of trust, safety, and reliability in computational systems. My research at ASU's Data Mining and Machine Learning Lab focuses on AI safety, how Large Language Models handle ambiguous inputs and how to make their outputs more trustworthy. My coursework in statistics helped me understand how uncertainty and bias in data is foundational to building secure systems. Growing up across multiple cities in India and then moving to United States for my undergrad gave me early exposure to how different communities experience security differently, from digital infrastructure access to physical safety. That awareness now shapes how I approach engineering problems, with attention to who is affected and how equitably solutions are distributed.

My coursework in Statistics complements my CS major by providing rigorous tools for quantifying uncertainty and identifying bias in data, which is essential for data security and trustworthy AI. Through GCSP, I aim to reduce the vulnerabilities and biases present in machine learning systems, contributing to AI that global communities can rely on.

My interest in security began during high school, where I served as Technology Captain of the Student Government and volunteered as a teacher at a school for students with special abilities. These early responsibilities taught me that technology serves people, and the people depending on it deserve systems that are reliable, fair, and secure. This principle guides my academic and professional trajectory.

MOTIVATION
& PURPOSE

BECOMING
A GC
SCHOLAR

Across the five GCSP competencies, my theme of Security has expanded well beyond its initial scope of digital security and encryption into a broader framework encompassing trust, safety, equity, and reliability.

Talent: My FURI research on ambiguity in Large Language Models addressed a core AI safety challenge, LLMs produce confident but incorrect outputs when processing ambiguous inputs, a reliability concern with direct security implications. This work resulted in an IEEE BigData 2024 publication (30+ citations) and confirmed my professional direction toward AI safety research.

Multidisciplinary: FSE 150 introduced the PESTLE framework, which reframed security as a challenge with political, economic, social, legal, and environmental dimensions. STS 330 extended this by examining how AI and data tracking reshape labor markets, democracies, and privacy whch are very real security concerns that transcend traditional cybersecurity.

Multicultural: HON 272 and ASB 378 provided historical-philosophical and contemporary-global lenses for understanding how different societies conceptualize trust and authority. This gave me crucial context for building solutions that serve diverse populations equitably.

Entrepreneurship: Venture Devils required translating technical capability into a viable product, testing whether Turing's privacy-first automation approach could meet real user needs beyond a research prototype.

Service: My EPICS project with Mayo Clinic on building an 'Autonomous Medical Resource Allocation System' grounded my understanding of security in healthcare resource allocation, where data security (HIPAA compliance) and health security (equitable ECMO access) are inseparable.

What ties these five competencies together is more than a shared theme label. Each experience built on the others: the PESTLE framework from FSE 150 shaped how I analyzed stakeholders in EPICS; the AI safety principles from FURI informed the privacy architecture in Turing; the multicultural lenses from HON 272 and ASB 378 changed how I think about who my technology serves; and the customer discovery discipline from Venture Devils taught me to validate assumptions before building. Security, as I now understand it, is not a single technical property but a system of trust, equity, and reliability that spans every dimension of engineering practice.

I am grateful to Professor Amy Trowbridge (FSE150), Professor Huan Liu of DMML, Professor Christiane Alcantara (HON272) and the whole GCSP community for shaping my understanding that effective engineering requires not only technical competence but also awareness of the social, cultural, and ethical systems in which technology operates.

contact@aryankeluskar.com
aryankeluskar.com