Natural Language Processing (NLP) in the medical field has the potential to dramatically influence the way in which everyday clinical care and medical research is conducted. NLP systems provide access to structured content embedded in raw medical texts, therefore enabling automated processing. There are however, several barriers prohibiting wide spread adoption of NLP technology primarily driven by the complexity and cost. This paper describes an approach and implementation which leverages cloud-based deployment and service-based interfaces to extract, process, synthesize, mine, compare/contrast, explore, and manage medical text data in a flexibly secure and scalable architecture. Through a virtual appliance architecture users are able to discover, deploy and utilize NLP engines on demand without requiring knowledge of the underlying, potentially complex, NLP engine. As highlighted in this paper, the system architecture can scale in several configurations: by increasing the number of instances deployed, the number of NLP engines, and the number of databases.