Deploying large-scale software systems often requires combining multiple different deployment technologies by creating heterogeneously nested deployment models, for example, using Kubernetes manifests within a Terraform configuration file. If such nested models need to be manually analyzed regarding the architecture of the final deployment, immense technical expertise in different technologies is required. Moreover, current automated architecture reconstruction approaches do not consider such heterogeneously nested deployment models. Therefore, we present a rule-based method that automatically transforms a given deployment model into a technology-agnostic, graph-based architecture model while considering heterogeneously nested deployment models. We applied the method in a case study to reconstruct the deployment architectures from six deployment models of three reference applications using different combinations of the technologies Ansible, Kubernetes, Terraform, Helm, Bash, and Docker. The case study shows that many architectural aspects can be reconstructed generically, but some parts need custom or semantic analysis transformation rules.