Option A is correct because DynamoDB uses cursor-style pagination through LastEvaluatedKey and ExclusiveStartKey. For an infinite scrolling feed, the application should request an initial page of results, then use the returned LastEvaluatedKey as the ExclusiveStartKey for the next query. AWS documentation explicitly states that the LastEvaluatedKey from a Query response should be used as the ExclusiveStartKey in the next Query request. This is the native DynamoDB pagination pattern and requires the least operational work. Full table scans to calculate offsets are inefficient and expensive at scale. Client-side pagination after retrieving all items is also inefficient and can break with large datasets. Storing page numbers manually recreates relational offset pagination and is fragile when feed data changes.

===============