Interview happened after take home assignment where I spent around 4 hours. The interview was scheduled based on my performace Two types of questions were asked: 1. How do you explain some technical concept to a business question? 2. How would you handle some scenarios as a consultant? Questions like 'how do you explain gradient boosting to business audience', 'how to you explain impact of change in coefficient of logistic regression' have no single acceptable answer. You are at the interviewer's whim on what amount of rigor or sugarcoating is required or the lack of it. Case Studies were vague and contrued at best. Here is an example: Consider, 'Company A has a DS team which has a built a model that does not have great performance. What would you do to improve the metric?' Me: ... feature engineering, model interpretability ideas ... emphasizing that one can get true understanding with data and business problem alone. Int: Why did you jump to modeling? Should you not understand the business problem? Me: We are talking about a hypothetical business problem and a data science solution thereof. If there was a model already built, likely business problem is clear. If I were to handle the problem with the client with multiple weeks of consulting time, I would start with business problem and understand the entire process. **It looked like you were looking out for improving the model specifically**. Had I said, 'I would look at the business problem', the interviewer could have said 'suggest a data science solution, assume that business problem is clear'. Data engineering conundrum Interviewer had problem with my CV having a few data engg terms. Having worked with startups and having led teams of DS/DE, you end up gaining knowledge on production tools, rest API's to serve your models and various trade-offs. Despite explaining this in great detail, interviewer asked me 'how do you describe yourself, more of ...'. ---- Finally, I gathered these points and decided to not proceed with further rounds: 1. With ambiguous questions, it looks like Bain will attract mostly fake data scientists who do more into 'talking in the air' and glamorizing datascience than concrete problem solving. Your C-level exec gives a damn about how xgboost/RNN works. 2. If they do not appreciate balanced skillset and require so called data scientist who imports a CSV file and creates a fancy presentation, please stop wasting precious time of people who take datascience seriously.