This workshop explores modern neural networks and pretrained foundation models for tabular data in R. Starting with a review of neural network concepts, the session introduces techniques such as residual connections and attention, then considers how pretrained models can be applied to familiar row-and-column datasets. Examples focus on the brulee and tabpfn packages, with TabICL discussed as an open-source alternative. Smaller datasets make the demonstrations accessible without requiring a GPU. Participants develop an understanding of the available modeling approaches, their computational demands, and the practical considerations involved in using these tools for structured-data analysis.