My PROD GPU Pod keeps on going off during traffic time. How to keep it awake 24/7 ?

Hi, I'm running FastAPI with some models with GPU. But there will be 0 traffic some-times, POD is going off and port gets changed. so when API Request comes it was unable to serve it back since FastAPI Port is killed & I've to reinstall packages + start server. How to keep the server up 24/7. Since POD runs on docker, I was not able to utilize systemctl to start/stop the App. any advice please ?? Thanks,