Add a Lambda GPU instance to your SSH configuration, then select it as a compute target in the local orx dashboard.
In the Lambda dashboard, open SSH Keys, click Add SSH key, and paste your computer's public key. Select that key when launching the instance.
cat ~/.ssh/id_ed25519.pub

Find the instance's IP address in Lambda. Instances use port 22 and the ubuntu user.
Host lambda-gpu HostName <ip-address> User ubuntu IdentityFile ~/.ssh/id_ed25519
ssh lambda-gpu
Run orx up, then open Settings > Compute > SSH. Test lambda-gpu and make it the default compute target if desired. The machine needs git and bash.