Four rounds.
Round 1 — Problem Solving (30 mins): partnered with a local grocery store to add a smart cart that lets users find that store's own products and prices on the smart cart, and at the same time see products and prices from other stores on the Instacart app through the same cart. Asked whether this is a good idea, what the hypothesis would be, which metrics to look at, and how to design the experiment.
Round 2 — Stats & Experimental Design (45 mins): a dozen or so rapid-fire questions, going from easy to hard. The easy ones were things like how to explain p-value to a PM, what the exact definition of p-value is, how to interpret a p-value, and how you'd decide if the p-value came out to 0.03.
The hard ones were things like how to interpret the individual parameters in a linear regression that has confounding factors, and how to use L1 and L2 regularized regression models.
Round 4 — Bar Raiser (60 mins): started with behavioral questions, then a case: on Sunday afternoons the number of orders accepted by shoppers dropped by two-thirds — why? This case was actually pretty complex, and you couldn't just run it through a framework the way Emma Ding's videos describe. You had to understand Instacart's product features and user pain points, lay out assumptions straight from the shopper/merchant/user angles for what the possibilities could be, and only then define the metrics. My suggestion: play it like a murder-mystery party game.
Round 4 — Project Review (45 mins): before the onsite they gave me a project, basically analyzing an A/B test on my own, and then in this onsite round I took the slides I'd prepared and gave a presentation.
Discussion
Loading comments…