Sustained attention and working memory are central cognitive processes that vary between individuals, fluctuate over time, and have consequences for life and health outcomes. Here we characterize the functional brain architecture of these abilities in 9-11-year-old children using models based on functional magnetic resonance imaging functional connectivity. Using data from the Adolescent Brain Cognitive Development (ABCD) Study, we asked whether connectome-based models built to predict sustained attention and working memory in adults generalize to capture inter- and intra-individual differences in sustained attention and working memory performance in youth. Results revealed that a predefined connectome-based model of sustained attention predicted children's performance on the 0-back task, an attentionally taxing low-working-memory-load task. A predefined connectome-based model of working memory, on the other hand, also predicted performance on the 2-back task, an attentionally taxing high-working-memory-load task. The sustained attention model's predictive power was comparable to that achieved when predicting adults' 0-back performance and by a connectome-based model of cognition defined in the ABCD sample itself. Finally, the working memory model predicted children's recognition memory for n-back task stimuli. Together these results demonstrate that connectome-based models of sustained attention and working memory generalize to youth, reflecting the functional architecture of these processes in the developing brain.