Due to continuously rising popularity, cloud and serverless computing became important buzzwords, causing many companies to consider transferring their systems to cloud-native architectures. With it the novel cloud service offering Function-as-a-Service (FaaS) plays an integral role in creating serverless architectures with the help of functions used as building blocks. However, making yourself dependent on just one cloud provider can lead to vendor lock-in problems. For this reason, it is important to diversify cloud providers and make serverless applications portable. This process requires developers in-depth technical expertise across many different cloud platforms, therefore making it error-prone and very tedious. This thesis elaborated a concept, which enables developers, lacking of this specific know-how, to model provider-agnostic workflow models of FaaS functions based on BPMN, which are used to generate provider-specific deployment packages. Additionally, the prototype BPMN2FaaS was implemented based on this concept, which is able to generate FaaS functions in the programming language Python and deployment packages supported by the platforms AWS Lambda and Microsoft Azure Functions.