Optical manipulation has greatly influenced biology, physics, optics, and nanotechnology. Fiber-based dual-beam traps are crucial for biological cell manipulation due to their flexibility and portability. However, achieving precise optically controlled 3D rotation of biological cells for optical tomography with isotropic resolution remains a challenge. This work presents a novel multi-core fiber-based dual-beam trap that enables program-controlled 3D cell rotation. A dedicated phase retrieval algorithm and a deep neural network based on a physical model facilitate complex wavefront shaping through the multi-core fiber, allowing real-time holographic control of the manipulation beam and enabling controlled rotation of human cancer cells in three dimensions. Conventional optical tomography is limited by the missing cone problem, which the proposed fiber-optic cell rotation addresses through multi-axis rotation, leading to accurate volumetric reconstruction with isotropic resolution. An autonomous tomographic reconstruction workflow utilizing computer vision and machine learning is implemented for robust 3D reconstruction of rotated cells. Additionally, precise quantitative phase imaging is achieved for the first time using an ultra-thin lensless endoscope. This research could pave the way for advancements in multi-core fiber light field control and imaging, expanding applications in clinical diagnostics, optogenetics, biosensors
