This study presents design, micro-fabrication and neural network calibration of a low cost micro-gas sensor. This sensor is able to identify two gases (CO and H2) and their concentrations in a mixture. Using the micro-fabrication process the sensitivity of the sensors has increased while decreasing the fabrication cost. Using the Pechini route technique, the sensing layer is fabricated based on a layer of metal oxide semiconductor (SnO2) which is doped with palladium. Sensing layer surface morphology has been characterized (using Differential Scanning Calorimetry (DSC), Scanning Electron Microscopy (SEM), and X-Ray Diffraction (XRD)) to study (i) the molecular structure of the sensing layer fabricated under various conditions, and (ii) the effect of palladium doping on the crystalline structure. Following our previous study, the neural network calibration is employed to perform identification process. Four different concentrations of Co and H2 are exposed to the fabricated sensor and the results of output voltages show the high sensitivity of the sensing layer.