Temporal single-cell transcriptomics enables the reconstruction of dynamic gene expression changes during development, yet its analytical power is often limited by data sparsity, technical noise, and imbalanced cell-type representation across time points. To overcome these challenges, we present GeneSys, a generative deep learning model that simulates single-cell transcriptomic landscapes under developmental constraints and informed by prior biological knowledge or user-defined hypotheses. GeneSys integrates a temporal variational autoencoder with a cell-type classifier and requires a lineage blueprint as input, allowing it to model the temporal transitions of transcriptional states with cell-type specificity. Leveraging data from Arabidopsis thaliana roots and mouse embryos, we show that GeneSys learns robust developmental trajectories, generates realistic and representative transcriptomes, and enhances gene prioritization tasks compared to unregularized scRNA-seq data.