mixed logit model using R

Hello, I am trying to estimate a mixed logit model on a panel dataset (for a company, up to five years observations were collected).
When I estimate the model and include random intercept for company (data are clustered at the company level), does this mean that I control for interdependence of firm-year observations at the company level? Secondly, i need to estimate the model on an estimation sample and validate it on the test sample. Will random intercept for company create a problem since most companies in the test sample are different from those from the estimation sample?