AWS Data Engineer case studies
AWS Data Engineer · Put your knowledge to work in context.
AWS Certified Data Engineer – Associate (commonly referenced as DEA-C01) is AWS's associate-level credential for engineers who build, operate, and maintain data pipelines on AWS. It sits between the foundational cloud certifications and the professional-level exams: candidates are expected to be comfortable with core AWS services and with SQL, and to reason about ingestion, transformation, storage, cataloging, orchestration, and access control as one connected system rather than as isolated products. The AWS Data Engineer case studies are 5 cases and about 3,894 words.
1Streaming Playback Events: Backlog, Memory Failures, and Small Files
A media analytics team ingests playback events through Kinesis Data Streams (48 shards, 2 KB average record, peak 6,000 records/s with a 3x diurnal swing). A Glue 4.0 streaming job with 10 G.1X workers, a 60-second microbatch, sessionization, and dimension lookups writes Parquet to s3://lake/events/ plus a full JSON copy to s3://lake/debug/.
838 words, about 4 min2Athena Over a Million Log Objects: Crawler, Projection, or Compaction
A SaaS company keeps 4.2 TB of ALB and CloudFront logs in s3://logs-prod/ under year=/month=/day=/hour=/, growing about 12 GB/day. Analysts query it with Athena from QuickSight; one 12-month dashboard query scans 310 GB and takes 4 minutes.
742 words, about 4 min3Overnight Chain: Duplicate Revenue, Worker Starvation, and a Cleared Sensor
A retail platform runs 130 Glue jobs, 22 Redshift stored procedures, and 40 dbt models on MWAA 2.7 with 3 x mw1.large workers. The chain starts at 01:00 and must finish by 06:00 for a 07:00 executive dashboard.
761 words, about 4 min4Claims Data: Persona Access, Region Filters, and Cross-Account Sharing
Account A owns a 3-year Parquet claims dataset in S3, registered in Lake Formation. Account B has 40 analysts in three personas.
794 words, about 4 min5Oracle CDC to S3: Redo Log Gaps, Duplicate Rows, and Change-File Flood
A financial services firm replicates an on-premises Oracle ERP (2.4 TB, 190 tables) to S3 using DMS full-load-plus-CDC into Parquet. Month-end redo volume hits 4,000 records/s for about three hours, against a baseline near 400.
759 words, about 4 min