Don't think you really know the field. On my team we almost exclusively use XGBoost or other boosted tree methods because it is typically the best model for tabular data. If we were working on CV or NLP that would be a different story and for that Neural Nets are by far the best models.