In order to access and utilise my machine learning models, I developed a custom API using Python with Nginx as the server. A Vue.JS frontend was applied to provide an interface. It was built using Docker containers to allow for easy deployment and scaling. The models are then in the source code of the server, which can be loaded when called upon. The API requires bearer tokens to access externally and pass data which is then used as the input to the model. The output is then sent as a response from the server.

FastAPI was used as the backend, as this is a lightweight implementation because the server does not require any complex business logic.