A transfer service uploads 20-GiB files to S3 over an unreliable WAN link. Single-request uploads are impractical, and retransmitting a whole file after a late failure is expensive. Which steps should the engineer implement? Select 2 answers.

1. Use a single PutObject operation and increase its payload above the single-upload size limit.
2. Treat an upload ID as a completed object and skip the CompleteMultipartUpload request.
3. Enable S3 Versioning to make failed network transfers resume automatically at the last byte.
4. Use multipart upload so individual failed parts can be retried without retransmitting successfully uploaded parts.
5. Add a lifecycle rule to abort incomplete multipart uploads after an appropriate number of days.