Instacart Machine Learning Engineer Interview Experience — An L4-to-L5 Change and Rejection

My process began with two coding problems, then ML fundamentals, linear regression details, and a brief delivery-time prediction design. The virtual onsite went deeper into search ranking, including cold start, pipeline components, models, and loss functions. The manager conversation covered motivation and past projects.

I had applied for L4 and expected one design round. I felt uncertain after that discussion, but the recruiter later gave positive feedback and said the L4 opening had disappeared. They added an advertising-system design round to assess me for L5. That interview covered the overall pipeline in detail. I felt confident afterward but was rejected without knowing why.

With four years of experience in Canada, I found myself awkwardly between level expectations. Looking back, understanding the full ML pipeline seemed more useful than concentrating on one topic such as calibration.