Acute myeloid leukemia (AML) is a heterogeneous hematopoietic malignancy with limited therapeutic options for many patients. Discoidin domain receptor 1 (DDR1), a transmembrane tyrosine kinase receptor, has been implicated in AML progression and represents a promising therapeutic target. In this study, we employed a hybrid virtual screening workflow that integrates deep learning-based binding affinity predictions with molecular docking techniques to identify potential DDR1 inhibitors. A multistage screening process involving PSICHIC, KarmaDock, Vina-GPU, and similarity-based scoring was conducted, leading to the selection of seven candidate compounds. The biological evaluation identified Compound 4 as a novel DDR1 inhibitor, demonstrating significant DDR1 inhibitory activity with an IC of 46.16 nM and a 99.86% inhibition rate against Z-138 cells at 10 μM. Molecular dynamics simulations and binding free energy calculations further validated the stability and strong binding interactions of Compound 4 with DDR1. This study highlights the utility of combining deep learning models with traditional molecular docking techniques to accelerate the discovery of potent and selective DDR1 inhibitors. The identified compounds hold promise for further development as targeted therapies for AML.