Web Services have become the de-facto standard for architecting and implementing business collaborations within and across organization boundaries. Web service composition refers to the creation of new (Web) services by combining the functionalities provided by existing ones. A process-oriented language for service composition has been proposed as BPEL4WS. BPEL4WS specification defines an XML based formal language and provides a general overview of the framework. However no design and implementation issues are described in it. Most of the available BPEL4WS compliant process engines are heavy weight, complex and not extensible. This paper describes the design and implementation of an embeddable, scalable and extensible BPEL4WS compliant process engine. This paper highlights the concepts and strategies that were followed during the design and implementation. Primary contribution of this paper is the design of stateless process model and the design of run time core engine using a multi-processor scheduler.