17.24Interview answer

“I'd first determine whether this should really be a full load or incremental sync. For large write volumes I'd use Salesforce Bulk API v2 rather than per-record connector calls. If Mule needs significant record-level validation, transformation, routing or restartability, I'd use a Batch Job and aggregate records before sending them downstream. I'd use External-ID-based upsert for replay safety, capture row-level failures, distinguish systemic from record-level failures, and reconcile counts/state after the job.”

Excellent.