Kaggle recently hosted a competition organized by the members of the CHemistry and Mathematics in Phase Space (CHAMPS) at the University of Bristol, Cardiff University, Imperial College and the University of Leeds to predict the magnetic interaction between two atoms in a molecule (i.e., the scalar coupling constant).  While it is possible to accurately calculate scalar coupling constants given only a 3D molecular structure as input using quantum mechanics calculations, these calculations are extremely expensive (days or weeks per molecule), and therefore have limited applicability in day-to-day workflows. We developed a state-of-the-art solution using deep neural networks which achieved second place amongst 2700+ competitors worldwide with a score of -3.22 (LMAE averaged amongst 8 j-coupling types) using an ensemble of 8 models, and the best single model performance across all teams with a score of -3.16 (LMAE), running inference in less than 20 milliseconds per molecule.