Title: From Calibration Plots to Agentic Quantum Control: NVIDIA Ising Calibration and QCalEval

Abstract:  Quantum processors require frequent calibration before they can run useful workloads. Today, this process depends heavily on experts interpreting experiment plots, judging calibration quality, and deciding on the next experiment. My talk will present NVIDIA Ising Calibration and QCalEval as a concrete example of agentic AI for scientific calibration workflows. NVIDIA Ising Calibration is a vision-language model designed to understand quantum calibration outputs and support agentic calibration. QCalEval is a benchmark for quantum calibration plot understanding, built from real quantum computer outputs across 22 experiment families and 87 scenario types. The talk will cover why calibration is a bottleneck for scalable quantum computing, how QCalEval evaluates scientific plot understanding, and how an AI agent can use Ising Calibration to help interpret results and recommend the next steps.