An Embedding is a vector representation of a data sample. The sample is as such "embedded" in the corresponding vector space. In NLP (natural language processing) applications, for example, words (or parts of words) need to be turned into embeddings in order to be processed. The embedding is often learned as part of ML modeling.