Research
Scientists still spend significant effort translating scientific intent into infrastructure actions. My research brings together AI agents, machine-learning models, and HPC systems to shorten that path while keeping scientific results inspectable and reproducible.
Scalable inference and HPC integration
How can many physics processes share accelerators effectively? I develop Inference as a Service infrastructure and its integration with scientific applications, including ATLAS Athena. This connects model execution to the practical questions of throughput, latency, concurrency, resource utilization, and deployment at HPC facilities.
Explore inference infrastructure
Physics applications
Particle physics supplies both the motivation and the tests for this work. My background includes Higgs boson studies with ATLAS, graph-based particle tracking, and generative simulation. Across these applications, computational performance must be evaluated alongside the quality of the scientific result.