Multivariate matching in observational studies typically has two goals: 1. to construct treated and control groups that have similar distribution of observed covariates and 2. to produce matched pairs or sets that are homogeneous in a few priority variables. This packages implements a network-flow-based method built around a tripartite graph that can simultaneously achieve both goals. The package also implements a template matching algorithm using a variant of the tripartite graph design. A brief description of the workflow and some examples are given in the vignette. A more elaborated tutorial can be found at < https://www.researchgate.net/publication/359513837_Tutorial_for_R_Package_match2C>.