How to fillna on new data after model deployment

For example, I have used fillna method to fill Age column(np.nan) records with Age-mean(i.e. 35) and Emp-Grade with mode-Grade(i.e. most frequent(mode): 5.2) and built a ML regression model and deployed it using flask web framework.

Independent features(X): Age, Experience, Gender, Grade and Dependent feature : Salary

User will enter required features and app will call predict API on submit, lets say the user dint fill Grade and kept as it np.nan, so we will have to fill grade with Grade-mode(i.e. 5.2) used in ML training, same is the case for Age.

So, are we going to store these Grade-mode, Age-mean(which are calculated during model training) in a config file under flask web app, so when a user do not provide Grade or Age, we will read config file and fillna respective np.nan value with Grade-mode, Age-mean?

and I have used IQR to update outliers values with upper boundary(Q3+1.5 IQR), lower boundary(Q1-1.5 IQR) instead of removing the outliers. Do I need to store these upper and lower boundary in config to update if a user enters Age, Grade outlier?

Any better approach instead of storing these values in config?