The convergence of Artificial Intelligence (AI) and the Internet of Things (IoT) offers significant opportunities for real-time, on-device intelligence. However, deploying computationally intensive models like Convolutional Neural Networks (CNNs) on resource-limited edge devices presents challenges related to energy, space, and latency. This paper introduces MITO AI MicroblazeV (Multiplier-Integrated Tiny Operator), a new hardware/software co-design that combines a RISC-V-based AMD MicroblazeV softcore processor with a custom-designed hardware accelerator. The MITO accelerator features a multi-stage pipeline, parallel processing units, and an innovative Radix-4 Booth multiplier to improve computational efficiency. It also uses an 8-bit integer quantization technique to decrease data size and resource use without significantly impacting model accuracy. Experimental testing of MITO AI MicroblazeV on a Xilinx Arty-A7 100T FPGA with the MNIST dataset shows that the system achieves a high classification accuracy of 97.54% and an impressive computational speedup of over 370x compared to a non-accelerated MicroblazeV system. The entire SoC consumes only 0.935W, making the MITO AI MicroblazeV a practical and energy-efficient platform for real-time edge AI in IoT applications.