This paper presents a novel embedded computing platform for 2D THz synthetic aperture radar (SAR) imaging targeting FPGA-based Systems-on-Chip. One of the key applications of the emerging THz SAR is an unmanned aerial vehicle (UAV) or mobile-robot based high-resolution environment mapping and profiling. Computational power along with energy consumption are the limiting factors with these mobile platforms. In this paper, a computing platform is proposed, which is suitable for stand-Alone UAV or mobile-robot based THz SAR sensing. A hardware/software co-design architecture is implemented using multiple hardware accelerators for the THz SAR compute-intensive kernels with a software-managed memory subsystem for large data size transfer between accelerators and external memory. The proposed hardware/software co-design is implemented and evaluated on a Xilinx Zynq MPSoC FPGA board. The experimental results are presented in terms of latency and energy consumption.