This project develops a dynamic pricing framework over a cloud based architecture, being scalable and highly configurable, considering the great cardinality of the solution in terms of the analytic models to build and apply. This architecture was defined using AWS and Terraform, ensuring an easy deployment agnostic to the client's infrastructure. The dynamic optimization of the prices is achieved by combining the training of a sales prediction model and the execution of a discount combination optimizer. The framework tries to be as general as possible in order to be easily adaptable to any given client. We provide general interfaces that can be reimplemented if the default implementations are not suitable for a given project. We performed simulations with data from a real client from the fashion retail sector, and the results obtained were promising, suggesting an improvement in the company's revenue.