This thesis focuses on the implementation of Near Infrared (NIR) calibration techniques for quality control of roasted coffee and the analysis of chemical variables influencing coffee brewing performance in capsules. The first aim of the research was the optimization of quality control during the roasting cycles by analyzing the color and moisture content of coffee. To obtain a calibration model for each feature, Partial Least Squares (PLS) regression with the use of spectral data preprocessed with first-derivative was chosen. The model was calibrated from the analysis of over 800 roasted coffee samples, characterized by different color and moisture levels. The tests were conducted at the Lavazza plant in Settimo Torinese, where this approach allowed for more efficient quality control on all roasting batches, reducing response times and increasing consistency in the final product.