Project
My research covers computer architecture, memory systems, and operating systems, with a focus on workload characterization. Current work spans energy-efficient and secure memory systems, artificial intelligence with a focus on diffusion models, and AI in computer science education.
Memory systems and computer architecture
DRAM subsystem research spans memory scheduling and refresh, protection against memory-level security vulnerabilities, and processing-in-memory architectures that compute inside DRAM. Address translation research focuses on TLB design for virtualized and GPU systems.
Artificial intelligence and diffusion models
Diffusion models are studied from the hardware up, including performance characterization of image, video, and 3D diffusion models across GPU architectures, activation sparsity in diffusion model layers, and training dynamics of temporal-attention heads in video diffusion transformers. Conditional diffusion models synthesize MRI from anatomical structure, lesions, and disease for neuroimaging applications. Large language model research targets memory-efficient serving.
AI in computer science education
AI tools for computer science education include multi-agent tutoring systems, AI-assisted quizzing, and knowledge-graph-based learning environments, designed and evaluated in real courses. Studies examine self-regulated learning, student disclosure of AI use, and assessment design when generative AI is widely available.