deploying, within a multi-tenant service provider network, an object storage location for an account associated with a user;

deploying, within the multi-tenant service provider network, a machine learning (ML) orchestrator associated with the account;

detecting, by the ML orchestrator, that a training dataset comprising a plurality of columns of training data has been stored at the object storage location;

determining a target variable to infer corresponding to a particular column of the plurality of columns based on an identifier of the particular column in the training dataset or based on a user selection of the particular column, wherein each column of the plurality of columns comprises a respective set of values;

identifying a plurality of ML pipelines to evaluate; wherein each ML pipeline of the plurality of ML pipelines comprises a respective set of one or more preprocessing operations to apply to the training dataset;

evaluating the plurality of ML pipelines including: (a) for each ML pipeline of the plurality of ML pipeline, applying the respective set of one or more preprocessing operations to the training dataset, and (b) training, using a ML training service of the multi-tenant service provider network, a plurality of ML models corresponding to the plurality of ML pipelines to infer the target variable, wherein the plurality of ML models are generated according to a plurality of ML algorithms of the plurality of ML pipelines;

deploying at least one of the plurality of ML models via a ML hosting service of the multi-tenant service provider network;

utilizing the at least one of the plurality of ML models to generate one or more inferences; and

storing the one or more inferences at the object storage location.