Runtime Attestation for Secure LLM Serving in Cloud-Native Trusted Execution Environments

Jianchang Su, Wei Zhang

Presentation: Virtual

Keywords: Trusted Execution Environment, Runtime Attestation, Large Language Models, Confidential Computing

Presenter Full Name: Jianchang Su

Presenter Email: [e-mail]

Abstract: Serving Large Language Models (LLMs) in cloud environments introduces significant security challenges, particularly protecting sensitive data from untrusted cloud components. While Trusted Execution Environments (TEEs) provide hardware isolation, current approaches offer only boot-time attestation without container orchestration integration. We present a framework addressing these limitations through: runtime attestation for continuous integrity verification, container-level measurement for multi-tenant environments, attestation-aware Kubernetes integration, and hardware-agnostic TEE abstraction. This comprehensive approach creates an unbroken chain of trust from hardware to application, enabling secure LLM deployment against both infrastructure and orchestration-level attacks while maintaining cross-platform compatibility.

Presenter Bio: Jianchang Su is a Ph.D. student at UConn, in the Department of Computer Science & Engineering. His research interests include cloud computing, serverless computing, and machine learning systems.

Paper Checklist Guidelines: I certify that all co-authors have validated the presented results and conclusions, and have read and commit to adhering to the Paper Checklist Guidelines, Call for Papers and Publication Ethics.

YouTube Link: https://youtu.be/n8CdNqTmrt4

YouTube Link Poster: NULL

Google Slides: https://docs.google.com/presentation/d/1cEBg0qHg6OO75eADFB6Uu21iCvMK07TJEG0ZkxOY_A8/edit?usp=sharing

Poster: No

Workshop Registration: Yes, the presenter has registered for the workshop.

YouTube Link Short: TBD

Submission Number: 16