A company has greatly increased the volume of CSV files stored in an Amazon S3 bucket. Data transformation scripts and queries have become much slower. An ML engineer must implement a solution that optimizes the data for query performance with the LEAST operational overhead. Which option satisfies this requirement?

Converting CSV files to Apache Parquet format using an AWS Glue ETL job is the most effective solution for optimizing query performance with the least operational overhead. Parquet is a columnar storage format that offers superior compression and encoding schemes, significantly reducing storage space and improving query speed by allowing engines to read only necessary columns. AWS Glue is a serverless ETL service, meaning it automatically provisions and manages the underlying infrastructure, minimizing operational overhead compared to managing an EMR cluster. Splitting CSV files into smaller objects might offer some parallelization but doesn't address the fundamental inefficiencies of row-oriented CSV for analytical queries. Dropping string columns is not a general solution for performance and could lead to data loss.