In terms of my own work, tarsnap is absolutely a work of engineering -- I very deliberately avoided doing anything new, instead using established and well-tested concepts. The exception to this is scrypt, which I designed -- and proved the security of -- because there was no existing password based key derivation function which met my standards for security. On that one occasion I crossed the line from engineering into science.
Science is great, but there's nothing to be ashamed about in doing engineering work. The world needs good engineers who can take basic scientific discoveries and make useful products out of them!