Fusion is not one-size-fits-all: Cross-Modal Representation Alignment for Time-to-Event Modeling | AIChainDay