Although large language models (LLMs) show promise for medical healthcare applications, their utility for personalized health monitoring using wearable device data remains underexplored. Here we introduce the Personal Wellbeing Coach Large Language Model, designed for providing well-being recommendations in food, fitness, sleep, and social-emotional health. The growing availability of wearable fitness tracking devices is increasing, and along with it, demand and new opportunities for personalized health monitoring. However, the application of LLMs for providing customized personalized well- being insights from multimodal wearable data remains unexplored. So this thesis presents the development of Personal Well-Being Coach LLM, a domain- adapted model designed to generate personalized recommendations in tasks like nutrition, fitness, sleep, and mental health.