Trustworthy Distributed AI
Georgia Tech VIP Team
About
Design and evaluate trustworthy AI systems that learn, reason, and act on sensitive data reliably and securely in distributed, adversarial environments. Aligned with the NAE Grand Challenge of securing cyberspace, the team asks: How can AI systems learn from sensitive data without exposing it? How can autonomous systems remain trustworthy under adversarial conditions? How can privacy, security, and utility be simultaneously guaranteed? Students will explore differential privacy, secure computation, federated learning, agentic AI, and robust distributed systems — targeting healthcare, genomics, finance, autonomous transportation, and critical infrastructure — producing researchers and engineers capable of building AI systems society can trust. The most valuable data in healthcare, genomics, finance, and critical infrastructure is also the most sensitive — and existing approaches like data anonymization are insufficient: high-dimensional and genomic data can be re-identified even after removing explicit identifiers. This team tackles the core challenge of building AI systems that are trustworthy by design: private (they do not expose individual data), secure (they remain correct under adversarial manipulation), and reliable (they behave predictably under failures or distribution shift). These three properties are deeply intertwined and often in tension with raw performance, making their simultaneous pursuit a rich and open scientific problem. As AI systems become increasingly agentic — reasoning, planning, and acting autonomously on behalf of individuals or institutions — new and more acute privacy and reliability challenges emerge, making this a rapidly expanding frontier with implications across healthcare, finance, autonomous transportation, and national security.
Majors
Algorithms, Combinatorics and Optimization, Computer Science, Cybersecurity, Bioinformatics, Biomedical Engineering, Computer Engineering, Electrical Engineering, Industrial Engineering, Machine Learning, Mathematics, Statistics
How to apply
Visit the official page to apply or learn more.