Simplify
Don't Over-Simplify

Liyi Zhou, Email: lzhou1110 (at) gmail (dot) com
[USyd] [Google Scholar] [Twitter] [Chinese]

Abstract

I am a Computer Science Lecturer (roughly equivalent to a US Assistant Professor) at the University of Sydney. I co-founded D23E.ch with Professor Arthur Gervais (UCL) and Dr Kaihua Qin (Yale).

My focus is on security and privacy issues, with the ultimate goal of developing generalized AI in system/software security to identify vulnerabilities, much like how Noam Brown created the first Poker AI. While my research spans beyond blockchain, I find it particularly engaging as a testbed due to its vast amount of open code and bytecode, transparent data from millions of users and bots, and real-world attacks involving billions of USD. Don't let the term "blockchain" mislead you though - my research in this space directly advances general system/software security principles.

I am Hiring!

I am actively looking for PhD students to join my research group at the University of Sydney. My primary vision is to develop the next generation of AI security agents - systems that can autonomously identify, analyze, and prevent security vulnerabilities, similar to how AI has revolutionized games like Poker. My research interests span across:
  • AI-powered security agents for vulnerability detection and prevention
  • Large security models - using LLMs to enhance existing security tools
  • Real-time intrusion detection and prevention 🔔
  • Advanced program analysis (fuzzing, symbolic execution)
  • Privacy-preserving transaction systems
PhD Candidates 🎓
If you are passionate about pursuing a PhD with me at the University of Sydney, please email me with:
  1. Three bullet points highlighting your achievements (anything)
  2. At least one technical achievement (research / engineering)
  3. Your CV

I am Seeking Funding Opportunities!

I am actively seeking funding and collaboration opportunities. My research philosophy emphasizes practical value and real-world impact - I believe the best research not only advances academic understanding but also solves concrete problems in industry. If you have interesting research questions you'd like me to explore, or if you're interested in my research and want to discuss potential opportunities, I'd love to hear from you. I'm always eager to tackle challenging problems that have real-world impact.

My Achievements

Rankings do not mean much, but they help illustrate my dedication to high-impact, top-tier research, especially for prospective PhD students.

My Research Summary

Click to expand research summary

Liyi is a researcher specialized in cybersecurity and artificial intelligence. Over the past five years, he has published extensively in leading international security conferences, and his work has been widely cited. His research mainly focuses on developing innovative ways to detect software vulnerabilities automatically, particularly by using advanced machine learning techniques. Liyi has built several practical, effective systems to identify security threats and anomalies in software. He has hands-on experience includes finding critical software vulnerabilities, which earned him multiple awards from bug bounty programs. He collaborates with top international researchers from universities like UC Berkeley and ETH Zurich, helping to extend the real-world impact and visibility of his research.

Liyi's current focus is about creating intelligent software tools that can independently identify and exploit software vulnerabilities without needing constant human supervision. His work introduces new ways for AI systems to reason about complex security issues on their own, adapting dynamically to emerging threats and iteratively refining their strategies. Using real-world data, especially from blockchain security incidents, Liyi's approach advances current cybersecurity methods through reinforcement learning, automated reasoning, and focused fine-tuning techniques. This research fills an important gap in existing technologies, moving cybersecurity systems from simple pattern matching towards more genuine, strategic problem-solving abilities. The goal is to significantly improve accuracy and reliability of automated cybersecurity solutions.

Liyi's research is valuable because it creates tools that can automatically identify software vulnerabilities, preventing dangerous exploits. These innovations can greatly benefit academic researchers, industries, governments, and broader society by reducing the need for costly and slow manual security reviews. Liyi's work aligns with Australia's national cybersecurity priorities, helping build local expertise in security technologies. Through mentoring and training the next generation of cybersecurity experts and through strong international collaborations, Liyi contributes to strengthening national cybersecurity capabilities. Additionally, insights from his research will shape responsible public policies around artificial intelligence, protecting important digital infrastructure and enhancing public trust in technology.

Liyi has demonstrated success previously by integrating artificial intelligence effectively with cybersecurity, shown clearly through practical prototypes and research projects he has completed. He receives strong support from the University of Sydney, including access to advanced computing equipment, funding resources, and expert collaborators in relevant fields. His project strategy focuses on adapting existing machine learning technologies and using cloud infrastructure to ensure his methods are practical and scalable. Initial experiments and prototype systems have already shown promising results, confirming the feasibility and potential of his approach.

My CV

Details of my publications, academic services, and experiences. [Download]