3 Stages. Stage 1 - cultural/experience interview with a few Python Questions to finish. Asked about tuples, lists, how they're stored in memory etc. Stage 2 - Take home test. Took a few hours. Implemented AdaBoost algorithm, SQL question, Python question using ByteIO to read to memory. Stage 3 - 2.5 hour long technical interview. Lots of questions about ML and CV particularly applied to their use-case (doorbell cameras). Questions included describing generalisation, precision/recall, handling imbalanced datasets, handling noisy data, etc.