We present an empirical model built on a high-resolution N-body dark matter simulation. We assume a redshift-independent star-formation efficiency for each halo to convert the accretion rate into a star-formation rate. Our model is calibrated using the z = 4 UV luminosity function (UVLF) and successfully predicts the observed UVLF at z = 5 – 10. We present predictions at z = 5 – 10 for UV luminosity and stellar mass functions, JWST number counts, the stellar-to-halo mass relation and star-formation histories. We combine this model with bleeding-edge reionization constraints (from z > 7 quasars, z ∼ 7 Ly α line-profiles, the updated Planck τ) to find new perspectives on the Epoch of Reionization (EoR). We find MUV < − 13.5 galaxies need an average fesc = 0.22 ± 0.05 to drive reionization and a highly compressed timeline: the IGM neutral fraction is [0.9, 0.5, 0.1] at z = [8.4 ± 0.2, 7.0 ± 0.2, 6.3 ± 0.2]. Inspired by the newly assembled sample of Lyman Continuum leakers that unanimously displays higher-than-average star-formation surface density (sigma), we fit a model tying fesc to sigma. Since sigma grows by > 2.5 dex over z = 0 – 8, our model explains the humble values of fesc at low-z. We find, strikingly, that < 5% of galaxies with MUV < − 18 account for > 80% of the reionization budget. We predict leakers like COLA1 (z = 6.6, MUV = − 21.5) become common towards the EoR and that the protagonists of reionization are not hiding across the faint-end of the luminosity function but are already known to us.