Turmeric, valued for its culinary and medicinal properties, has increasingly become vulnerable to harmful adulteration, posing serious health risks. Among these, ferrous sulfate heptahydrate is particularly concerning due to its potential to cause iron overload and oxidative stress. This study intended to develop a rapid, cost-effective, and non-destructive technique for finding such adulteration using Fourier Transform Infrared (FTIR) spectroscopy in Attenuated Total Reflectance (ATR) mode. Fresh turmeric samples collected from five locations in Tamil Nadu, India, were mixed with ferrous sulfate heptahydrate at concentrations of 0%, 5%, 10%, 15%, 20%, 25%, 30%, and 100%, yielding a total of 36 samples. FTIR spectra revealed distinct peaks for turmeric (1630 cm⁻1, 1745 cm⁻1, 2930 cm⁻1, 3720 cm⁻1) and for the adulterant (1060 cm⁻1), providing clear spectral markers for adulteration detection. The spectral dataset was analyzed using Multiple linear regression (MLR), random forest (RF), K-nearest neighbors (KNN), eXtreme Gradient Boosting (XGBoost), and Artificial neural networks (ANN). Among these, XGBoost achieved the peak overall performance, through an R2 of 0.9968, mean squared error (MSE) of 0.0007, root mean squared error (RMSE) of 0.0265 (dimensionless), and very low mean absolute error (MAE). ANN also produced comparable accuracy, confirming the robustness of nonlinear approaches. These findings demonstrate that combining FTIR with advanced machine learning, particularly XGBoost, provides a powerful framework for rapid, reliable, and non-destructive adulteration detection in turmeric, with potential applicability to other food powders.