I applied online. The process took 3 weeks. I interviewed at Instacart (San Francisco, CA) in Jul 2018
Interview
1. SQL test
2.Recruiter screen
3. Data take home challenge
4. Phone screen with a data scientist
5. On site
Overall it was a smooth interview process. The recruiter was very helpful. The data scientists and PM's I met seemed smart but also very serious (could sense some stress). The on site has a behavioral competent (freebie if you can communicate decently) as well as several technical interviews. They will ask about statistical techniques you would use to solve a specific scenario (monte carlo simulation, decision tree, etc.). The interview was pretty rigorous and they will gauge how advanced you are with statistical knowledge and ability to apply it to business problems. Overall I thought I did well in all interviews other than a portion of one of them (didn't know what statistical method to use to solve specific case question), and ended up getting rejected.
Interview questions [1]
Question 1
1. We are observing this trend, what are possible explanations for why?
2. Behavioral (talk about a time when...)
3. SQL. Did not have live coding, only in the initial problem and take home portion
Faced a deep technical challenge during the interview, asked to design a personalization algorithm for product recommendations. I walked them through feature engineering and model choices, which stressed me out initially. Luckily, I realized the structure was almost identical to a problem I had explored on prachub.com while prepping. The interview progressed through a behavioral round where they focused on my previous experiences, and I ultimately received an offer, but decided to decline after careful consideration. Overall, it was intense but rewarding.
It was pretty straightforward. HR and then technicals. The technicals were a mix of statistics and math questions. Overall, it moved pretty quickly. My interviewer was quite late with no apology and in general not very friendly. I knew someone interviewing at the same time and they had a much better experience and recieved the same questions.
Interview questions [1]
Question 1
Suppose we wanted to launch 15-minute deliveries in a specific area where we already deliver. How would you statistically test the results?
