My advice is, go directly to the source and read the papers of the services you are using (dynamo, s3, etc.). IMO doesn't make sense to become an "AWS expert".

My advice is, go directly to the source and read the papers of the services you are using (dynamo, s3, etc.). IMO doesn't make sense to become an "AWS expert". I would take an engineer that can compare SQS to Kafka (self-managed vs kinesis), MQQT and RabbitMQ and offer an educated guess on which is better suited on our use case, taking into consideration the context (no of engineers, skillset required, application characteristics, etc.) than someone who's going to answer "SQS!" when a service requires a queuing system.