Corrugated Web Steel Beams (CWSB) have gained significant recognition in industry and academia over the past few decades. The unique design of corrugated webs has brought forth numerous advantages, such as enabling the creation of lightweight and thin structures. Furthermore, it has enhanced the beam web's shear buckling strength and stability, reducing the need for transverse stiffeners, commonly required for flat webs. Several parameters have been identified as influential factors in the shear buckling strength of CWSB, highlighting the complexity of understanding this phenomenon. This complexity emphasizes the need for a reliable model that can accurately predict the shear strength of CWSB. In this study, Machine Learning (ML) techniques were employed to investigate and predict the shear strength of CWSB. Traditional methods like Linear Regression and Tree Models to advanced techniques like Gaussian Process Regression and Support Vector Machines (Quadratic) learning methodologies are explore. Due to its inherent adaptability and performance, there is particular emphasis on exploring Neural Network Models, particularly the Optimizable Neural Network Model. An array of new experimental data was gathered from various sources, including 206 trapezoidal and 65 sinusoidal tested Corrugated Web Steel Beam specimens. ML algorithms were then utilized to develop prediction models, revealing the relationships between the principal response parameter (shear strength of CWSB), various input geometric parameters (web length, web depth, width of longitudinal fold panel, longitudinal width of inclined fold panel, corrugation depth, width of inclined fold panel, and web thickness), and material properties (steel yield strength). Among the developed models, the Optimizable Neural Network emerged as an exemplary predictor of the shear strength for sinusoidal and trapezoidal steel corrugations. It exhibited exceptional performance with an R2 value of 0.96, a mean relative error of 4%, and a coefficient of variation (cov) of 3%,