Zilin Shen

I am a Ph.D. candidate in Computer Science at Purdue University, advised by Prof. Elisa Bertino. My research focuses on protocol security, formal verification, and LLM security.

I am also a Founding Engineer at Eigen AI, where I work on LLM infrastructure, including serving optimization, CUDA kernel development, and production deployment.

Previously, I received my B.E. from Tsinghua University in 2022.

Experience

Founding Engineer, Eigen AI, Aug 2024 - Present
- LLM serving infrastructure and optimization
- CUDA kernel development (FP4 Attention for B200)
- Production deployment with Kubernetes and Kong Gateway
Network Security Research Intern, Cisco, Summer 2024

Research Interests

LLM Systems: Inference optimization, serving infrastructure, kernel development
AI Security: LLM safety, protocol security (MCP), formal verification
Network Security: Wireless protocol analysis, fuzzing

Education

Ph.D. in Computer Science, Purdue University, 2022 - Present
B.E. in Software Engineering, Tsinghua University, 2018 - 2022

selected publications

NDSS

WCDCAnalyzer: Scalable Security Analysis of Wi-Fi Certified Device Connectivity Protocols

Zilin Shen, Imtiaz Karim, and Elisa Bertino

In Proceedings of the 33rd Network and Distributed System Security Symposium (NDSS), 2026

Acceptance Rate: 16%

Abs

We present WCDCAnalyzer, a scalable approach for security analysis of Wi-Fi Certified Device Connectivity protocols.
USENIX

AKMA+: Security and Privacy-Enhanced and Standard-Compatible AKMA for 5G Communication

Yang Yang, Guomin Yang, Yingjiu Li, and 9 more authors

In Proceedings of the 34th USENIX Security Symposium (USENIX Security), 2025

Acceptance Rate: 17%

Abs

We propose AKMA+, a security and privacy-enhanced authentication and key management architecture that is compatible with 5G standards.
NeurIPS

PSMBench: A Benchmark and Dataset for Evaluating LLMs Extraction of Protocol State Machines from RFC Specifications

Zilin Shen, Xinyu Luo, Imtiaz Karim, and 1 more author

In Proceedings of the 39th Annual Conference on Neural Information Processing Systems (NeurIPS), 2025

Acceptance Rate: 24.5%

Abs

We introduce PSMBench, a benchmark and dataset for evaluating large language models on extracting protocol state machines from RFC specifications.
AsiaCCS

Segment-Based Formal Verification of WiFi Fragmentation and Power Save Mode

Zilin Shen, Imtiaz Karim, and Elisa Bertino

In Proceedings of the 19th ACM Asia Conference on Computer and Communications Security (AsiaCCS), 2024

Acceptance Rate: 21%

Abs PDF

We present a segment-based formal verification approach for analyzing WiFi fragmentation and power save mode protocols, identifying security vulnerabilities in protocol design and implementation.
Preprint

Proving DNSSEC Correctness: A Formal Approach to Secure Domain Name Resolution

Qifan Zhang, Zilin Shen, Imtiaz Karim, and 2 more authors

arXiv preprint, 2025

Under Review

Abs

We present a formal approach to proving DNSSEC correctness for secure domain name resolution.
MobiCom

Combating Link Dynamics for Reliable LoRa Connection in Urban Settings

Shuai Tong, Zilin Shen, Yunhao Liu, and 1 more author

In Proceedings of the 27th Annual International Conference on Mobile Computing and Networking (MobiCom), 2021

Abs PDF

We propose techniques to combat link dynamics for reliable LoRa connections in urban environments, addressing challenges in long-range low-power wireless communication.