View of Performance‑Aware Fine‑Tuning and Inference Optimization for Large Language Model Code Generation: A Unified Framework | Stanford Database Library of American Journal of Applied Science and Technology