In the present paper, we propose a methodology for ensuring on-demand individualized job training to a learner via a virtual organization (VO). We offer suggestions for the development of a platform supporting the lifecycle of such a job training VO. The proposed methodology addresses three stages in assisting the learner: (1) helping the learner identify his/her personalized learning path (in terms of curricular requirements, or competences) for a specific job, (2) searching for the Pareto optimal combinations of instructors to cover the identified competences, and (3) building a VO as a holonic multi-agent system (HMAS) composed of candidate resources, in an attempt to partially bridge the gap between VOs and VO environments.