"Trained immunity" describes long-term functional reprogramming of innate immune cells that enhances responses to secondary challenge. However, the extent to which diverse training stimuli converge on shared transcriptional programs remains unclear. Here, we systematically analyzed publicly available RNA-sequencing datasets from diverse human trained-immunity models to define both baseline and restimulated responses of trained cells. Cross-dataset comparisons identified shared transcriptional and functional programs, indicative of common baseline and trained immune states. By integrating gene-expression data across models, we derived a consensus signature of human trained-immunity, which was validated in independent bulk and single-cell datasets. Enrichment analyses showed activation of pro-inflammatory and metabolic programs, including modulation of cellular iron homeostasis and robust increases in chemokine CCL7 expression, alongside suppression of anti-inflammatory, resolution-associated genes. These findings indicate that diverse training stimuli converge on core, shared gene-expression programs and support a two-layer model in which durable baseline reprogramming precedes exaggerated inflammatory responses upon restimulation.