Dear Editors, We are pleased to submit our survey, “Mapping the Neuro-Symbolic AI Landscape by Architectures: A Handbook on Augmenting Deep Learning Through Symbolic Reasoning,” for consideration at the Neurosymbolic AI Journal. Neuro-symbolic AI has seen substantial progress in recent years, but the growing diversity of approaches can make it difficult to understand how the different frameworks relate to one another, what their respective strengths and limitations are, and how they can be used in practice. In this survey, we propose a systematic mapping of logic-based neuro-symbolic approaches according to their architectures. In addition to providing a broad overview of the field, our goal is to offer a practical perspective: in particular, we focus on frameworks that allow practitioners to augment existing neural models with symbolic knowledge while treating the symbolic component as a black box. A central component of the survey is a fairly detailed set of preliminaries on probability, logic, and statistical relational learning (SRL). We have deliberately devoted substantial space to this material because these formalisms are not merely background for the subsequent discussion: the operations and inference mechanisms they provide, e.g. SAT, abduction, MAP inference, and probabilistic logic programming, are fundamental to our architectural classification. Understanding how these operations work, and their computational properties, is important for understanding why different neuro-symbolic architectures exhibit different strengths, limitations, and scalability characteristics. We therefore use detailed, progressively developed examples to provide readers who are primarily familiar with machine learning with sufficient grounding to understand the architectural distinctions made later in the survey. Our selection of the literature likewise reflects the purpose of the survey. This work is not intended to be a snapshot cataloguing the most recent papers in neuro-symbolic AI. Rather, we aim to identify and explain the overarching architectural ideas that organize the field, and to demonstrate that the broad range of existing approaches can be meaningfully situated within the proposed taxonomy. Consequently, we focus particularly on high-level ideas of the different architectures found in the space and deduce their strengths and weaknesses.For each architectural family, we discuss a small number of representative frameworks in more detail, using them to illustrate how the underlying ideas are implemented, focusing on the papers that introduced the different concepts rather than attempting to provide an exhaustive catalogue of individual papers. We believe that this perspective can complement existing surveys of neuro-symbolic AI by providing both a conceptual map of the field and a practical handbook for researchers and ML practitioners interested in incorporating symbolic reasoning into neural systems. Thank you for considering our submission. We look forward to hearing from you.