Imputer function limitations

I want to know that the imputer function replaces all the null values in the dataset with the strategy type defined in the function. eg : mean/median / fixed value etc.
If for eg there are multiple columns having null values and according to our study, one column mean best fits the null value, in other median best fits and in some a fixed value best fits. Then whether imputer function work in that case . and if yes, how?

You can make two different imputer instances, then you can fit and transform those two instances seperately. Choose the columns which fits median for one and some other strategy for other. I hope it helps.