Developing Machine Learning applications may require a reasonable knowledge of infrastructure and how to properly provision it from the software developer. In this paper, Machine Learning applications are exploited using Serverless infrastructure resources available on cloud computing platforms. Thus, it is proposed to analyze its benefits, difficulties and limitations, based mainly on the point of view of the developer. From this perspective, the process of developing Machine Learning applications becomes something simpler, thus helping the developer deliver software of this type faster, more efficiently and provisioned.