Aminoglycosides are an important class of antibiotics that play a critical role in the treatment of life-threatening infections, but their use is limited by their toxicity. In fact, gentamicin causes severe nephrotoxicity in 17% of hospitalized patients. The kidney proximal tubule is particularly vulnerable to drug-induced nephrotoxicity due to its role in drug transport. In this work, we developed a perfused vascularized model of human kidney tubuloids integrated with tissue-embedded microsensors that track the metabolic dynamics of aminoglycoside-induced renal toxicity in real time. Our model shows that gentamicin disrupts proximal tubule polarity at concentrations 20-fold below its TC50, leading to a 3.2-fold increase in glucose uptake, and reverse TCA cycle flux culminating in a 40-fold increase in lipid accumulation. Blocking glucose reabsorption using the SGLT2 inhibitor empagliflozin significantly reduced gentamicin toxicity by 10-fold. These results demonstrate the utility of sensor-integrated kidney-on-chip platforms to rapidly identify new metabolic mechanisms that may underly adverse drug reactions. The results should improve our ability to modulate the toxicity of novel aminoglycosides.