The upcoming Prime Focus Spectrograph on Subaru PFS will push our understanding of the cosmology and galaxy evolution to higher redshifts with unprecedented depth. In this talk, I will provide an overview of recent developments in numerical and machine learning tools to analyze this data. I’ll review TARDIS, the forward modeling maximum a posteriori framework for reconstructing large scale structure from joint high redshift (z~2.5) probes. I’ll show application of this technique to the recent data release from the CLAMATO Lyman Alpha Forest Survey. Using a generative deep learning method, we can also marginalizing over hydrodynamic uncertainties of the intergalactic medium at this redshift. Time permitting, I will also discuss extending this technique to lower redshift ranges by folding in photometric redshift uncertainties within the likelihood function.