There are a growing number of underwater applications, including climate change monitoring, marine biology research, oil rigs exploration, unmanned operations, search and rescue, underwater navigation, and scuba diving. Most of these applications demand reliable, flexible, and high-speed underwater sensing and communication systems. Despite significant advancements in terrestrial and space communication, high-speed underwater wireless communication remains in its infancy due to the harsh environmental conditions, unique signal propagation challenges, and a lack of infrastructure. The most widely used underwater communication methods - acoustic, radio frequency (RF), and optical waves - each face trade-offs. Acoustic signals can travel long distances but suffer from low data rates and high latency. RF signals offer higher data rates but are significantly attenuated in water, limiting their effective range to just a few meters. Optical communication holds great promise for delivering high-speed data transmission, however, it remains underutilized in underwater systems due to issues such as light scattering, absorption, misalignment, and sensitivity to environmental disturbances. To address these limitations, this project aims to develop an AI-defined high-speed underwater optical networking system that integrates sensing and communication into a unified architecture. In such a system, sensing and communication mutually enhance each other: real-time sensing informs more effect