In their pursuit of market competitiveness and sustainable top line growth, enterprises are increasingly turning to sophisticated analytics solutions to derive insights and value from the deluge of data that are being generated from all sources. Leading practitioners of Big Data analytics have already moved past the stage of using single analytics modalities on siloed data sources. They are starting to create composite analytics solutions that take advantage of multiple analytics programming models and are also integrating them into their existing enterprise IT systems. At the same time, the CIOs have wholeheartedly embraced cloud computing as a means of reducing the capital and operational cost of their IT systems and streamlining their DevOps processes. Platform-as-a-Service (PaaS) as a cloud computing consumption model has seen wide acceptance by developers and IT administrators. Although there are PaaS platforms for individual workload types involved in these advanced composite analytics solutions, the composition aspect is not addressed by any of these individual PaaS platforms. Further, there is no lifecycle management support for the solution as a single logical entity. This paper argues for the need of a true PaaS for composite analytics solutions in order to accelerate their adoption by the industry and foster the creation of a healthy ecosystem. We present the design and prototype implementation of such a platform and our early experience of using it to deploy a Telco Fraud Detection solution.