Abstract NASA's ICESat‐2 (Ice, Cloud and land Elevation Satellite‐2) satellite launched in 2018, carrying a single instrument, the Advanced Topographic Laser Altimeter System (ATLAS). The Level 1 science objectives of the mission focus primarily on the cryosphere, with specific interest in monitoring changes in polar ice sheets, glaciers and sea ice. However, in addition to planned observations and data products for polar, land, vegetation, ocean and the atmosphere, ATLAS's photon‐counting, green‐wavelength instrumentation enables impressive bathymetric measurement capability. Most of the ICESat‐2 along‐track data products were developed during pre‐launch studies, without a dedicated effort focused on bathymetry. The absence of a dedicated bathymetry product has required the scientific community to develop independent, individual algorithms for bathymetric signal extraction, most often tailored to local or regional studies. No existing approaches have been proven applicable to global application. Over the last 3 years, the ICESat‐2 Project Science Office has sought to address the need for coastal and nearshore bathymetry through the development of a Level 3a, along‐track data product for global shallow‐water bathymetry (ATL24). The ATL24 workflow embraces several independent signal extraction algorithms in a machine learning ensemble to provide robust signal extraction of the sea floor and sea surface heights in variable environmental conditions and water quality. This paper explains the approach to the algorithms and an assessment of the algorithm performance to evaluate the usefulness for high‐priority science and application use cases.