Galaxies’ distribution and their redshift evolution are essential elements in cosmological research. The Analysis of Resolved Remnants of Accreted Galaxies as a Key Instrument for Halo Surveys (ARRAKIHS) is an upcoming European Space Agency mission designed to advance our understanding of the Universe. ARRAKIHS aims to image faint extended halos of nearby galaxies, providing insight into the nature of dark matter. Moreover, the wide field of view and unprecedented flux sensitivity (31 magnitude/square arcsec) of ARRAKIHS give us a unique opportunity to constrain cosmological parameters using dN/dz, the redshift-dependent differential number density of background galaxies. This study utilizes the dataset from COSMOS2020, which is a comprehensive multi-wavelength photometry and photometric and spectroscopic redshift catalog of approximately 1.7 million sources across a 2-square-degree area of the sky. We transform this dataset into the ARRAKIHS framework, using its total throughput for each photometric band. Using machine learning techniques to predict redshifts from transformed photometric data, ARRAKIHS’ potential to provide accurate, reliable measurements of galaxy redshift distributions is assessed. This advancement in measuring galaxy redshift distributions will significantly deepen our understanding of the universe, especially by testing the Λ Cold Dark Matter cosmological model and the role of dark matter in galaxy formation. The preliminary Fisher-matrix forecasts indicate that an ARRAKIHS-like dN/dz analysis can achieve uncertainties of approximately σ(Ωₘ) ≈ 0.05 and σ(w) ≈ 0.3 with σ(z) ≈ 0.07(1+z) at z < 1.0, which translates to 95% confidence intervals of ΔΩₘ ≈ ±0.1 and Δw ≈ ±0.6 even in the pessimistic case.