This work presents a fast and uncertainty-aware sequential data assimilation framework suitable for estimation of key aerodynamic states (e.g., instantaneous vorticity fields and aerodynamic loads) during severe gust encounters, where vortex–gust interactions strongly affect the flow dynamics. The framework comprises an ensemble Kalman filter (EnKF), designed to detect and reconstruct nearly-impulsive flow disturbances with a wide range of strengths and orientations and introduced at arbitrary times. The forecast and measurement update (analysis) stages of the EnKF are each composed of learned operators in a low-dimensional latent space, obtained via a physics-augmented autoencoder. The forecast operator propagates the undisturbed baseline dynamics, but cannot predict random gust-induced deviations from these dynamics. Thus, the analysis stage frequently assimilates new surface pressure measurements to listen for disturbance signals and initiate deviation from the nominal trajectory. The methodology is trained and tested on flowfield snapshots from high-fidelity simulations of two-dimensional airfoil-gust encounters and corresponding sparse surface pressure data. Because assimilation occurs entirely within the reduced-order latent space, the updates are computationally efficient and ensure that aerodynamic states can be continuously estimated from streaming pressure data. The estimated instantaneous latent state remains physically interpretable via decoding to the original high-dimensional flow. Eigenvalue decomposition of the state and observation Gramians in the measurement update stage reveals the dominant correction directions necessary to capture the flow disturbance and quantifies how sensors inform the state corrections throughout gust interaction. The framework readily accounts for sensor failure; sensor dropout experiments show that the EnKF adaptively re-weights neighboring sensors to compensate for lost information, preserving estimation quality even under degraded sensing configurations.