This paper presents simulations of the metastable helium (He * ) observations of WASP-107b, so far the highest signal-to-noise ratio detection that is confirmed by three different instruments. We employ full 3D hydrodynamics coupled with coevolving nonequilibrium thermochemistry and ray-tracing radiation, predicting mass-loss rates, temperature profiles, and synthetic He * line profiles and light curves from first principles. We find that a stellar wind stronger than solar is demanded by the observed heavily blueshifted line profile and asymmetric transit light curve. Radiation pressure can be important for Lyα observations, but not He * . Our model finds that WASP-107b is losing mass at a rate of ´-Å -M M 1.0 10 yr 9 1 . Although M varies by <1% given constant wind and irradiation from the host, shear instabilities still emerge from wind impacts, producing ∼10% fluctuations of He * transit depths over hour-long timescales. The common assumption that He * transit depth indicates the fluctuation of M is problematic. The trailing tail is more susceptible than planet adjacency to the shear instabilities; thus, the line profile is more variable in the blueshifted wing, while the transit light curve is more variable after midtransit. We stress that the synergy between Lyα (higher altitudes, lower density) and He * (lower altitudes, higher density) transit observations, particularly simultaneous ones, yields better understanding of planetary outflows and stellar wind properties.