Pathogenic bacteria are harmful microorganisms capable of causing diseases. To fight or eliminate those microorganisms, antibiotics with antimicrobial action have been developed wich can be synthetic or semi synthetic. Through time, bacteria has developed mechanisms to fight this antimicrobial action, generating the antibiotic resistance. This issue is a serious global problem that affects the health area. Because of this, a workflow based on the KDD methodology. The proposed approach use data obtained through MALDI-TOF mass spectrometry techniques without preprocessing, in conjunction with deep learning, to implement a multiclass classification neural network of bacteria at the species level, with the purpose of obtaining a fast and reliable recognition. Different tests were implemented to this neural network, obtaining promising precision results with the aproach, reaching accuracy 99.15% the highest and 98.09% the lowest. Implementing evaluation metrics such as confusion matrix and classification reports to measure the recognition of species at the individual level, finding cases in wich the precision was 100% the highest and 94% the lowest.