With billions of objects on the horizon of LSST first light, researchers will need new ways to detect and measure sources. Among the challenges of maximizing the science available to us with this volume of data is the estimation of object redshifts, due to the impossibility of spectroscopic follow-up of all extragalactic sources. Photometric redshift (photo-z) estimation is thus a top priority of LSST as weak lensing, SN cosmology, galaxy/AGN evolution, and more all require robust measurements of redshift. I will give an overview of the photo-z landscape, current and new paradigms for photo-z estimation, and present work on developing DeepDISC photo-z, an image-based photo-z estimator trained and validated on simulated LSST data. I will contextualize our results with traditional photo-z methods and discuss the advantages and limitations of our pz estimator, along with avenues of future application.
