Data Science and Predictive Modelling Intern

Two months, one question

Which internship listings actually get filled, and can you predict it on the day the listing is posted? That is the question this two month project exists to answer, and nobody on the team knows yet.

Your working week

You start in pandas with a listings table covering the last eighteen months, join it against application and hire outcomes, and find out quickly how much of it is missing. From there you build a classifier, probably logistic regression first because you will need something interpretable to argue with, then a tree ensemble to see how much signal you left behind. Feature engineering is the heart of it: posting day, skill count, title length, how specifically the location is described. You present coefficients to the operations team, they push back, and you go and check.

Practicalities and what you take away

Three days a week in the Delhi NCR office, two from home. This role is unpaid. In exchange you get a genuine outcome dataset rather than a teaching set, a named mentor from the data team, and a project you can walk somebody through in any interview. Final year students and recent graduates only.