As the prevalence of age-related diseases rises, understanding and modulating the aging process is becoming a priority. Transcriptomic aging clocks (TACs) hold great promise for this endeavor, yet most are hampered by platform or tissue specificity and limited accessibility. Here, we introduce Pasta, a robust and broadly applicable TAC based on a novel age-shift learning strategy. Pasta accurately predicts relative age from bulk, single-cell, and microarray data, capturing senescent and stem-like cellular states through signatures enriched in p53 and DNA damage response pathways. Its predictions correlate with tumor grade and patient survival, underscoring clinical relevance. Applied to the CMAP L1000 dataset, Pasta identified known and novel age-modulatory compounds and genetic perturbations, and highlighted mitochondrial translation and mRNA splicing as key determinants of the cellular propensity for aging and rejuvenation, respectively. Supporting Pasta\'s predictive power, we validated pralatrexate as a potent senescence inducer and piperlongumine as a rejuvenating agent. Strikingly, chemotherapy drugs were highly enriched among pro-aging hits. Taken together, Pasta represents a powerful and generalizable tool for aging research and therapeutic discovery, distributed as an easy-to-use R package on GitHub.