A company operates a data lake in Amazon S3 that stores large datasets in multiple formats. The company has an application that retrieves and processes subsets of data from multiple objects in the data lake based on filtering criteria. For each data query, the application currently downloads the entire S3 object and performs transformations. The current process requires a large amount of transformation time. The company wants a solution that will give the application the ability to query and filter directly on S3 objects without downloading the objects. Which solution will meet these requirements?