I have a xgboost model file, now I have 250 million rows data to predict scores using this model, how can I use hive-udf to do this , so I can add jar in hive

I just saw someone transforming xgb model to sql in python ,get the result just like this
case when (f2354 is null or f2354 >= 1393.75) then when (f4534 is null or f4534 >= 280.5) then when (f5718 is null or f5718 >= 263)
…
but it is too long ,
and Unfortunately my java is not too bad ..