About
I am the John W. Hancock Jr. Professor of Engineering at Virginia Tech. My home department is Computer Science, and I hold a courtesy appointment in Electrical and Computer Engineering. I direct the PEARL laboratory and serve as associate director of the Stack@CS Center for Computer Systems.
My research concerns how multiprocessors and accelerators are programmed and orchestrated. This work spans high-performance computing, cloud computing, and artificial intelligence. It has informed the OpenMP standard and is used in Red Hat Enterprise Linux and commercial compilers.
My current work focuses on efficient and resilient AI systems across the computing continuum. I also study performance engineering for modern accelerators and systems for data-intensive science. A common thread runs through all three: reducing the infrastructure footprint of AI.
I am a Fellow of the IEEE, a Royal Society Wolfson Fellow, and a Distinguished Member of the ACM. I chair the Steering Committee of the ACM International Conference on Supercomputing. Before returning to Virginia Tech, I was Head of the School of Electronics, Electrical Engineering and Computer Science at Queen’s University Belfast. I received my Ph.D. from the University of Patras in 2000.
Selected recent work
- “Intent-Aware Caching for Efficient LLM Serving.” NeurIPS, 2026, to appear.
- “WISP: Waste- and Interference-Suppressed Distributed Speculative LLM Serving at the Edge.” ACM SIGMETRICS, 2026.
- “APEX: Asynchronous Parallel CPU-GPU Execution for Online LLM Inference on Constrained GPUs.” IEEE IPDPS, 2026.
- “dLLM-Serve: Bridging the Memory Gap in Diffusion Language Model Serving.” ACM ICS, 2026.
- “HiRED: Attention-Guided Token Dropping for Efficient Inference of High-Resolution Vision-Language Models.” AAAI, 2025.
- “ParvaGPU: Efficient Spatial GPU Sharing for Large-Scale DNN Inference in Cloud Environments.” SC, 2024.
All publications · Google Scholar · DBLP · CV · Short bio