Xipeng Shen
North Carolina State University, NC·Computer Science
Computing now happens mostly in warehouse-scale datacenters, on hardware that is increasingly varied and shared among many users, and on datasets so large that simply storing and moving them is a bottleneck. Projects in this lab attack both sides of that squeeze: how to keep cloud computation private and efficient when the machines are heterogeneous and the provider cannot be fully trusted, and how to shrink scientific data so drastically that it can be used without ever being unpacked.
- Fields
Computer Science
Engineering
Education Research
- Topics
trusted execution environment
secure enclave
cloud security
- Grants
- 4 known grants, through Apr 2030
What they work on
- Design a data abstraction for Trusted Execution Environments that lets data exist and be managed separately from the enclave computing on it, addressing weak inter-enclave sharing, absent intra-enclave isolation, and poor scalability.
- Extend enclave protection to CPU-GPU collaboration, multi-socket systems, and disaggregated or pooled memory by resolving the missing data abstraction in current secure enclaves.
- Develop EPOD, a lossless compression format whose data can be accessed and manipulated directly without decompression, and extend its coverage to floating-point scientific datasets.
- Build multi-level EPOD operation support as a library plus a scalable synthesizer based on large language models, and add continuous compression so the format fits streaming workflows at scientific facilities.
- Create serverless computing abstractions that exploit the distinct hardware characteristics of heterogeneous datacenters while strengthening the security of the environment serverless functions run in.
- Integrate cyberinfrastructure, artificial intelligence, and environmental science training into nine existing graduate courses and a two-week summer bootcamp, supported by an online platform offering personalized hands-on exercises and progress tracking.
More about the research
A fourth effort builds training programs so that environmental scientists and computer scientists can actually use this kind of infrastructure.
Why it matters
More and more computation and sensitive data move to the cloud, so better enclave designs could let a broader set of applications run with protection from provider-side vulnerabilities, improving security and privacy for cloud users generally. Processing compressed scientific data directly could cut dataset sizes by one or two orders of magnitude without quality loss and speed analysis, easing storage costs at scientific facilities. The training project prepares students to apply AI and large-scale computing to coastal flooding, hurricanes, and marine ecosystem change.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Collaborative Research: SaTC: CORE: Medium: Data Enclave: A Data-Centric Trusted Execution Environment
NSF · May 2026 – Apr 2030
Collaborative Research: SHF: Medium: EPOD: Data Reduction for Science through Efficient Processing Without Decompression
NSF · Jul 2025 – Jun 2028
Collaborative Research: CSR: Medium: Scaling Secure Serverless Computing on Heterogeneous Datacenters
NSF · Oct 2023 – Mar 2028
Closest labs
Chengming Zhang
University of Houston, TX·Computer Science
Adapt GPU error-bounded lossy compressors such as cuSZ to real scientific workflows, supporting diverse data formats and compression quality targets, with quality checked by tools like QCAT.
Radu Sion
Stony Brook University, NY·Computer Science
Design plausibly deniable storage systems that hide the existence of data from a multi-snapshot attacker, one who inspects a device repeatedly, while holding the user's devices and credentials.
Peng Jiang
University of Iowa, IA·Computer Science·Early-career
Develop a runtime system that gives a virtualized view of data partitioned across heterogeneous devices, plus a compiler tool that places data using a statistical performance model of random memory accesses.
The school
North Carolina State University: 40 labs on PI Pages, $717M a year on research, 603 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.