AWS Certified Machine Learning Engineer Associate (MLA-C01) - 340 Questions

1. A machine learning engineer needs to organize raw training datasets in Amazon S3 so that SageMaker training jobs can efficiently load only the data belonging to a specific experiment without scanning the entire bucket. Which S3 design practice best achieves this?

2. A data engineering team stores ML training files in Amazon S3 using the Parquet format. They want to ensure SageMaker training jobs can read the data as fast as possible and reduce Amazon S3 API costs. Which S3 access pattern should they implement?

3. A machine learning team has raw CSV files landing in Amazon S3 every hour. They need to make those files queryable via standard SQL for ad-hoc analysis without writing custom ETL code. Which AWS service combination is the MOST appropriate?

4. A company wants to ingest real-time clickstream events into a data lake on Amazon S3 for downstream ML model training. The events arrive at up to 50,000 records per second and must be stored in Parquet format with no custom servers to manage. Which AWS service should they use?

5. A machine learning engineer needs to reuse precomputed features across multiple SageMaker training jobs to avoid redundant feature computation. The features must be retrieved with low latency (< 10 ms) during online inference as well. Which AWS service should the engineer use?