Primary resistance to tyrosine kinase inhibitors (TKI) is a significant barrier to optimal outcomes in chronic myeloid leukemia, but little is known about the factors contributing to response heterogeneity. Using scRNA-sequencing, we identified eight statistically significant features in pretreatment bone marrow mononuclear cells which correlated with either sensitivity (major molecular response or MMR) or extreme resistance to imatinib (eventual blast crisis transformation). Employing machine-learning, we also identified LSC and NK gene expression profiles predicting imatinib response with >80% accuracy, including zero false positives for predicting BC. A canonical erythroid-specifying (TAL1/KLF1/GATA1) regulon was a hallmark of LSCs from patients with MMR and was associated with erythroid progenitor (ERP) expansion in vivo (p<0.05), and a marked 2-10-fold (6.3-fold in Group A vs 1.09-fold in Group C) erythroid over myeloid bias in vitro. Notably, ERPs demonstrated exquisite TKI sensitivity compared to myeloid progenitors (p<0.001). These LSC features were lost with progressive resistance, and in patients who transformed, MYC- and IRF1-driven inflammatory regulons became evident. Patients with MMR also exhibited a 56-fold expansion (p<0.01) of a normally rare subset of hyperfunctional adaptive-like NK cells (CD57+NKG2C+) which diminished with progressive resistance, while patients destined for BC accumulated inhibitory NKG2A+ NK cells favoring NK cell tolerance (through HLA-E binding on target cells). Finally, we developed a parsimonious set of antibodies to validate our scRNA-seq findings. This panel will be useful in prospective studies of primary resistance, and assessing the contribution of predetermined versus acquired factors in TKI response heterogeneity.