Physics-integrated inference for signal recovery in non-Gaussian regimes

High-performance room-temperature sensing is often limited by non-stationary 1/f fluctuations and non-Gaussian stochasticity. In spintronic devices, thermally activated Néel switching creates heavy-tailed noise that masks weak signals, defeating linear filters optimized for Gaussian statistics. Here, we introduce a physics-integrated inference framework that decouples signal morphology from stochastic transients using a hierarchical 1D CNN-GRU topology. By learning the temporal signatures of Néel relaxation, this architecture reduces the Noise Equivalent Differential Temperature (NEDT) of spintronic Poisson bolometers by a factor of six (from 233.78 to 40.44 mK), effectively elevating room-temperature sensitivity toward cryogenic limits. We demonstrate the framework's universality across the electromagnetic and biological spectrum, achieving a fivefold error suppression in Radar tracking (relative to raw data), an 83% uncertainty reduction in LiDAR, and achieving a 15.56 dB SNR in ECG (a 5.8 dB enhancement over raw transduction). This hardware-inference coupling recovers deterministic signals from fluctuation-dominated regimes, enabling near-ideal detection limits in noisy edge environments.