The rapid evolution of labor markets, particularly in emerging economies like Cambodia, necessitates a dynamic and real-time skill demand analysis. Traditional methods, such as employer surveys, are often static and infrequent, failing to capture the fluidity of an evolving workforce. This research addresses this data gap by presenting a novel computational pipeline that leverages recent advancements in Large Language Models (LLMs) and Natural Language Processing (NLP). The proposed framework integrates multilingual text processing, LLM-driven information extraction, and taxonomy-based skill alignment using internationally recognized standards such as the European Skills, Competences, Qualifications and Occupations (ESCO) and the World Economic Forum (WEF) job taxonomy. Evaluation results demonstrate strong extraction performance for key job fields, while skill extraction shows high conceptual alignment despite lower strict-match accuracy. The findings confirm the potential of LLM-based methods to enhance labor market intelligence and provide actionable insights for policy, education, and workforce development in Cambodia.