Ambient noise tomography (ANT) has become a powerful technique for imaging the Earth’s subsurface by utilizing the seismic noise recorded by seismometers worldwide. However, to successfully implement ANT, certain requirements must be met, such as sufficient data length and dense sensor network with a high-performance computer to do the calculation. In this study, a robust and computationally efficient ANT algorithm is presented. It is designed to determine subsurface velocity models with minimal computational resources. To validate the effectiveness of the algorithm, synthetic models that incorporate unique source characteristics, serving as a controlled environment for evaluating the inversion process are developed. These synthetic tests demonstrate the algorithm’s ability to manage complex source configurations and produce accurate and detailed velocity models. The ANT algorithm is applied to real-world seismic data from two geologically distinct regions: South California and the Appalachian Mountains. The resulting velocity models provide valuable insights into the subsurface structure and properties of these areas, revealing intricate geological features and anomalies that have important implications for understanding the region’s tectonic history and present-day dynamics. The successful application of the ANT algorithm to both synthetic and real-world data demonstrates its potential for a wide range of geophysical applications, including geothermal exploration, oil and gas prospecting, and geohazard assessment. This study represents a significant advancement in the field of subsurface imaging, offering a robust and efficient tool for unraveling the complexities of the Earth’s interior.