Patrick Watters
Building scalable, resource-efficient AI systems for high-performance computing
I am a Ph.D. candidate at the University of Nevada, Reno working on systems for modern AI workloads in high-performance and distributed computing environments.
My research focuses on the infrastructure behind large-scale AI — from training and inference to memory, data movement, scheduling, and resource management — with the goal of making increasingly demanding workloads more efficient and adaptable.
I am especially interested in emerging systems challenges in agentic AI, efficient LLM serving, KV-cache management, and zero-knowledge machine learning.
Current research
LLM inference and serving
Designing adaptive and resource-aware inference systems for dynamic, heterogeneous compute environments.
ML systems and infrastructure
Improving data movement, scheduling, and resource utilization for large-scale training and inference workloads.
Distributed & verifiable AI systems
Exploring scalable coordination and execution for distributed and verifiable machine learning workloads.