Xiangyu Zhang

Xiangyu Zhang is the Samuel D. Conte Professor of Computer Science at Purdue University. He specializes in AI security, software analysis, and cyber forensics, with a focus on detecting vulnerabilities in traditional software and AI systems. His research includes dynamic slicing, program profiling, and securing AI models against backdoors. He has led projects funded by DARPA, IARPA, NSF, and industry, securing over $13 million in grants. He has mentored 30+ PhD students and postdocs, many of whom hold academic positions or industry roles. His work has been recognized with awards like the ACM SIGPLAN Distinguished Dissertation Award and NSF Career Award.

Education:

• PhD, Computer Science, University of Arizona (2006)
• MS & BS, Computer Science, University of Science and Technology of China (2000 & 1998)

Research Interests:

• AI Security: backdoor detection, adversarial attacks, model integrity
• Program Analysis: dynamic slicing, static/dynamic debugging
• Cyber Forensics: log analysis, attack investigation
• Software Reliability: fault localization, vulnerability detection

Key Projects:

• DARPA V-SPELLS: domain-specific program analysis
• IARPA TrojAI: AI backdoor detection competitions
• NSF-funded research on model debugging and co-reasoning of software/NL artifacts

Awards & Recognition:

• Top performer in IARPA TrojAI competitions (13/18 rounds)
• ACM SIGPLAN Distinguished Dissertation Award (2006)
• NSF Career Award (2009)
• Ranked 8th globally in systems research productivity (2024)

Career Contributions:

• Secured over $13M in research funding
• Guided 25 PhDs and 8 postdocs
• Published 89+ top-tier papers (2014–2024)