In this book chapter, we will discuss the methods for developing advanced data analytic frameworks for ocean wave prediction and vision-based underwater sensing. Accurate prediction of ocean wave patterns is crucial for applications ranging from coastal management and maritime navigation to offshore energy harvesting and climate modeling. We will explore how machine learning and deep learning algorithms can be trained on multi-modal datasets: including satellite observations and data collected from our custom-built bio-inspired underwater robot, Aquabot, and meteorological inputs to predict short-term and long-term wave behaviors with improved accuracy and spatial resolution. The chapter will also cover recent advancements in vision-based underwater sensing technologies, including on-device machine learning models for underwater environmental monitoring and exploration.