Exploring the world beneath the ocean has been a difficult task, especially in depths that are unsafe for humans. Studying underwater creatures can be very sensitive due to disturbances resulting from typical remotely operated vehicles (ROVs) currently used. Soft robots consist of elastomeric materials, compliant actuators, sensors and other supporting structures, which enable them to be used for numerous applications due to their flexibility, light weight, low noise, and many degrees of freedom. Mechanical actuators, such as pneumatic actuators and servomotors, introduce design constraints related to their size, weight, and cost. Moreover, vibration and noise are undesired attributes that preclude the use of robots developed with such technologies and might disturb the test environment. This paper presents a robust design of a jellyfish-like robot with eight bell segments, named KryptoJelly. The presented robot can perform multidirectional swimming by NiTi shape memory alloys (SMA) actuators confined in a conduit and activated by electrical current stimulation. KryptoJelly is constructed from a 3D printed rigid structure and a soft silicone bell that closely mimics the biological locomotion and appearance of a jellyfish found in nature; the Chrysaora colorata species. Four 127 μm diameter SMA wires (of mass 14 mg each) were used per channel to deform the silicone bell margin. The robot can operate up to 1608 cycles continuously for 1.5 h underwater at high power input (3 times higher than the standard) and sustain its own total body mass of 650 g (∼6000 times the weight of the actuators used). KryptoJelly is able to maneuver in both vertical and horizontal directions during bell contraction-expansion cycles. A study on the effect of multistage-power-time input sequence of NiTi SMA actuators and bell design, which results in swimming, is presented. This work has shown the great potential of employing smart materials in biomimetic soft robots, that can be deployed for eco-friendly underwater exploration or other applications.