Recent advances in convolutional neural networks have shown promise for a wide range of engineering applications including production quality assessment. This has been enabled in part by the availability of open-source algorithms and techniques such as transfer learning. Despite these factors, deploying these algorithms in production environments remains a challenge. The research presents an architecture for the deployment of convolutional neural network algorithms for the quality assessment of additive manufacturing (AM) build processes. The first iteration of this architecture has been implemented in a preproduction powder bed fusion AM facility at the National Centre for Additive Manufacturing in the UK. By demonstrating the application of this architecture on data generated from in-process monitoring data, the study hopes to reduce barriers faced when taking machine learning models from a laboratory concept to a production facility. By reviewing the latency of predictions, the study highlights that the additional proposed architectures may be viable for processing live production data for detection and control of defects. Finally, the study proposes that AM build machine manufacturers could provide a service-oriented architecture to encourage the implementation of in-process correction strategies.