The cosmic large-scale structure (LSS) provides a unique testing ground for probing gravity, dark energy, inflation, dark matter, and the neutrino-mass scale. The coming decade will bring an unprecedented influx of data from LSS surveys, including DESI, Euclid, LSST, and SPHEREx. To fully exploit this wealth of information, we must move beyond traditional summary statistics and harness the full statistical power of field-level inference. This workshop will focus on the advantages of field-level inference, the theoretical and computational challenges involved, and its application to real and simulated data. By working directly with full map-level data, we can extract significantly more cosmological information, break degeneracies, and improve constraints on fundamental physics.