The article proposes an approach based on the use of knowledge graphs (KG), integrating basic fundamental and applied industry knowledge in order to create a supporting cluster for a specialist (beginner or experienced) in solving specific engineering or more broadly - business problems. The graph of fundamental mathematical engineering knowledge, formed on universal approaches and principles for KG, can serve as the basis for the formation of a local knowledge cluster necessary for solutions to an applied engineering problem (as, possibly, for problems of a wider profile). To do this, an algorithm for layer-by-layer clustering of the graph of fundamental mathematical knowledge is proposed. Graph structure design is based on two key ideas: the nesting of topics on the principle from particular to general and the substantive relationship between sections. Topics as elements of fundamental mathematical knowledge were taken as tops. The algorithm takes three key parameters as input: the set of target vertices, the set of familiar vertices. The output of the algorithm is a local cluster of vertices, the minimum necessary for mastering the target vertex-themes (the development route is individualized for a specific query). The graph and the layer-by-layer clustering algorithm can perform general didactic goals, and also allow for parallel development of projects, problem solving and unpacking of missing elements of knowledge.