Robust Optimization for Service-Level-Constrained Resource Allocation in Large Language Model Inference Serving. JMTA 2026, 11 (4s), 756-767.