Data Science Case Study and Defence

What they ask

The data science track at the AI and analytics hub screens with a case study rather than a coding test. The sequence reported for an intern position: apply in summer, hear back about three months later with a case study assessment to be returned within one week, then another long wait, then a ninety minute interview built around your submission. Elapsed time from application to interview: about six months, so do not plan around it.

The interview has no fixed script and runs in three parts:

1. Background. Your degree, the projects you have done (the graduation project gets the most time) and the technologies you used in each.
2. Presentation. You walk through your case study solution end to end: how you framed the problem, how you cleaned and explored the data, which model you chose, how you evaluated it.
3. Questions. Nearly all of them start with "why". Why this technique and not a simpler one? Why this library? Why that metric? How does the method actually work underneath? A few basic data science questions (bias and variance, overfitting, leakage between train and test) are mixed in.

What they look for

That you own every choice in your notebook. The interviewers care less about the final score than about whether you can justify the pipeline, name the alternatives you rejected and explain the mechanics of the method you picked. Prepare the graduation project to the same depth; it is probed with the same "why" questions, and the technologies you name there become the next question.