A Coding Implementation on Microsoft’s Phi-4-Mini for Quantized Inference Reasoning Tool Use RAG and LoRA Fine-Tuning