Cloud interview questions for data engineers
AWS, GCP, and Azure storage, compute, and warehouses, plus IaC, scaling, and cost. 81 questions with full answers, written for freshers. All topics
AWS (9)
- Amazon S3 and storage classes
- AWS Glue vs Amazon EMR
- Amazon Redshift (and vs Snowflake)
- Amazon Kinesis (and vs Kafka)
- IAM roles, execution policies, and least privilege for AWS data pipelines
- S3 data lake partitioning, prefix design, and small files
- Cost governance in AWS: Athena, Glue, and Redshift
- VPC Endpoints (Gateway vs Interface) and KMS for S3 data lakes
- Athena partition projection
GCP (2)
Azure (2)
General Cloud (1)
GCP in Depth (10)
- Pub/Sub delivery, ordering and dead letters
- Dataflow autoscaling and Streaming Engine
- Dataflow vs Dataproc
- Bigtable vs BigQuery vs Spanner vs Cloud SQL
- GCS consistency and object storage behavior
- Datastream and CDC on GCP
- Dataplex and governance on GCP
- GCP IAM service accounts, Workload Identity, and pipeline security
- Cost governance in GCP: BigQuery queries and storage
- Private Google Access, VPC Service Controls, and CMEK on GCP
AWS & Azure (7)
Cloud Architecture Basics (8)
Azure in Depth (14)
- Azure Entra ID and RBAC for data pipelines
- ADLS Gen2 Medallion directory layout and lifecycle management
- Controlling costs in Azure Synapse and Data Factory
- Azure Synapse Dedicated SQL Pool table distribution
- Azure Data Factory pipeline triggers and Integration Runtimes
- Azure Event Hubs partitions and Event Hubs Capture
- Securing Azure data pipelines with VNets, Private Endpoints, and Key Vault
- Choosing between ADF, Synapse Pipelines, and Databricks on Azure
- Querying data lake files with Azure Synapse Serverless SQL
- Azure Data Factory building blocks: pipelines, activities, datasets, linked services
- Incremental load in Azure Data Factory with a watermark
- Metadata-driven ADF pipelines with Lookup and ForEach
- CI/CD for Azure Data Factory: Git, ARM templates and environments
- Error handling in Azure Data Factory: retries, failure paths and pipeline status
AWS Core Services (10)
- AWS Lake Formation permissions and cell-level security
- AWS Lambda VPC networking and private RDS access in ETL
- Amazon Kinesis Data Firehose buffering and lake ingestion
- AWS Step Functions distributed map for high-concurrency ETL
- Cross-account IAM delegation and ExternalId security in data lakes
- AWS Glue Streaming versus Managed Service for Apache Flink
- AWS PrivateLink and VPC Interface Endpoints for data exfiltration defense
- Active-active disaster recovery with S3 Cross-Region Replication
- Amazon EMR Serverless capacity pre-initialization and cost tuning
- S3 Object Lock and immutable compliance storage in data platforms
Azure Data Engineering (7)
- Azure Data Factory Self-Hosted Integration Runtime architecture
- Azure Event Hubs capture to ADLS Gen2 and consumer groups
- Azure Private Endpoints and virtual network security for ADLS Gen2
- Azure Managed Identities and Key Vault secrets elimination
- Azure Databricks Delta Lake optimization: liquid clustering and predictive I/O
- Metadata-driven ingestion framework in Azure Data Factory
- Multi-region disaster recovery for Azure data lakes with paired regions
Google Cloud Data Platform (11)
- Google Cloud Storage storage classes and lifecycle automation
- Google Cloud Pub/Sub ordering keys and exactly-once delivery
- Cloud Dataflow windowing and late data handling with watermarks
- Dataproc Serverless for Apache Spark batch pipelines
- Cloud Composer environment architecture and Airflow worker autoscaling
- VPC Service Controls and preventing data exfiltration in GCP
- Cloud Dataflow stateful processing and streaming autoscaling
- Workload Identity Federation for keyless CI/CD in GCP pipelines
- Pub/Sub to BigQuery direct streaming ingestion without Dataflow
- Multi-region streaming resilience with Pub/Sub and dual-region GCS
- Automated sensitive data inspection and tokenization with Cloud DLP