Latency-Aware Resource Scheduling for Large Language Model Inference Systems. JDIAS 2026, 1 (3).