The question is moot because the Linux OS and firmware running on the Raspberry Pi can just keylog your private key anyway, or transparently use its own when it comes time to actually generating your public key, or choose its own nonce values when signing, or any number of things. You either trust it or you don't. And when you're known to trust it then you can become a target of whoever makes it. That's why Trezor makes their own hardware. https://wiki.trezor.io/Security

How do you input/output transactions with this device? How do you back up your keys?

I hate to be a downer though, so here's some comic relief https://dilbert.com/strip/2001-10-25