From Cloud to Kathmandu: Architecting Quantized LLM Inference on the Edge for Nepali Businesses Have you ever felt the friction between the immense potential of large language models and the stark realities of deploying them in resource-cons…