First round in an OA with two parts. A theory part with general linear regression questions (ridge, lasso, successive orthogonalization), and a data part where you need to find use a linear model to fit some data that is heteroskedastic, and find a way to practically deal with heteroskedasticity and adjust estimator variances. After the OA is passed there a first round of interview (pair programming). Then a second round, followed by a final "virtual on-site".