Fraud detection in financial environments is a critical area whose objective is to identify anomalous patterns or activities that may indicate fraudulent practices. A financial fraud detection system with a web interface was designed using supervised models, such as Random Forest (RF), Logistic Regression (LR) and Support Vector Machine (SVM), as well as unsupervised models, such as Isolation Forest (IForest), Local Outlier Factor (LOF) and One Class Support Vector Machine (OCSVM). The study used the evolutionary incremental methodology composed of the spiral and CRISP-DM with a quantitative and explanatory approach, to apply machine learning techniques and measure performance by evaluating the models, for the development of the interface Streamlit was used in a virtual environment, which includes modules for login, detection and reporting of results. In particular, the supervised SVM model presented an accuracy rate of 99%, and the unsupervised LOF model with 97%, both models show a good ability to identify positive cases of fraud.