You’ll have infrastructure deployment engineers no matter what you do. But you’ll only need infra ops staff (as opposed to application ops staff) if you’re running your own infra.
There’s nobody you pay, in your own company, to sit around watching S3 or Lambda or SQS for degradation. That ops service is part of AWS’s job that you pay for through each service’s fee structure.
