The performance of Electric vehicles (EVs) depends on the conditions like terrains, sudden start/stop, and other uncertain scenarios which create an imbalance between load torque and electromagnetic torque. Due to this imbalance, severe torsional vibrations occur which causes an uncomforting situation for passengers and put extra mechanical stress on EVs. The sensorless, reactive power-based Model Reference Adaptive System (Q-MRAS) is suitable for precise speed estimation and control but in low-speed regions and with uneven load conditions, its performance degrades. To mitigate these issues and make the system robust against parametric variations, external disturbance, and operational uncertainties, an improved Model Reference Adaptive System based Integral Sliding Mode Control (MRAS-ISMC) is proposed. The MRAS integrated with the ISMC makes the drive simply robust against parametric variation by balancing the load torque and electromagnetic torque. The proposed MRAS-ISMC topology is verified on various periodic, non-periodic load torque disturbances and varying reference speeds.