The model combines feature windows, sequence embeddings, and threshold calibration for low false-positive rates.

windowed = telemetry.rolling(window=24).mean().dropna()
alerts = model.predict(windowed) > 0.78
P(fault|x) = sigma(Wx + b)