The Experience API (xAPI) has the potential to collect all manner of data associated with the experience of a learning event. This experience includes not only the traditional content of the learning objective, but also can include information about the learning environment, the learner (physiological, social (including other learners), psychological, etc.) and potentially about the instructor, delivery method and other environmental factors. As implementation of the xAPI grows, the amount of data available increases. This data, when provided with enough breadth and depth can be useful in understanding what comprises an ideal learning experience. That experience will be unique to each learner and can even vary among different instances of learning for a single learner. Traditional learning environments (i.e. one teacher to many students) cannot possibly provide the highly personalized learning experience that can otherwise be enabled by the xAPI due to physical limitations faced by a single instructor. A natural evolution of learning spaces that can take advantage of the personalization information provided by xAPI are immersive virtual environments (IVEs) where virtual advisors and instructors can be utilized. While there has been some discussion on the limitations of learning in virtual worlds, they are less related to content delivery than they are about comfort and cultural acceptance. That notwithstanding, IVEs (virtual worlds) provide an ideal location/platform to provide highly personalized learning experiences. This paper discusses how avatars can be created and personalized based on a learners likes and dislikes and how, with certain design criteria, can be created to maximize factors such as trust and influence. This paper is a first step in defining what data could be useful when collected for the creation of a personalized learning experience, as well as defining a process for the design and the deployment of an avatar based virtual advisor.