I applied through a recruiter. The process took 2 weeks. I interviewed at Capital One
Interview
I was contacted by a recruiter on LinkedIn. She was absolutely the best part of the process. The interview consists of four parts: a hacker rank challenge, a phone screen with a current data scientist, a data set challenge, and an on-site interview. If you feel nervous about the hacker rank challenge, you can practice on other areas of the site. The test involves dynamic programming, natural language processing, some combinatorics and a simple regression problem. I found some of the wording confusing but comparing your answers to the test cases is helpful. You won't get to see all the test cases.
The recruiter contacted me within an hour of submitting the test to let me know I passed to the next round. The phone screen was scheduled very quickly for a conference call. The interviewer never showed up to the conference call though. I let the recruiter know after waiting on hold for 15 minutes and she quickly resolved the problem. The nice thing about their phone interview is that there are no trick questions (think: how many ping pong balls fit in a 747?). Most of the questions are open-ended: if you were given this data set, what do you think the possible problems are? How confident would you be in your answers? There is one map-reduce question if you have experience with that. There will also be some behavioral questions (use the STAR technique to answer).
The recruiter contacted me later in the day to let me know I had a very positive response from the interviewer. The on-site interviews are set on a fixed schedule (Fridays every two weeks). She gave me the option of taking the data test right away or waiting a week because I had some scheduling conflicts. I was told that there were still a few slots available to interview and that if I finished within a few days I might be able to get one. The data set challenge is one basic machine learning question on an anonymous data set and then a descriptive analysis plus an open-ended question about baby names. They say it's designed to take about 4 hours but that you can take as much time as you feel you need. You're supposed to wait 2 to 3 business days to hear back.
The recruiter contacted me at the end of 2 business days. She had been delaying e-mailing me because she thought she would have had news earlier. She said she would get back to me at the latest the next morning. A day passed and I didn't hear anything. Then I got the bad news that I didn't get the interview.
The biggest highlight of the whole process was the recruiter. She's really fantastic at her job.
Interview questions [1]
Question 1
If you're trying to predict the gender of your customers and you only have 100 data points, what are possible problems?
I applied online. The process took 1+ week. I interviewed at Capital One (Richmond, VA) in Dec 2014
Interview
1. A recruiter called who said they mainly hire Ph.D's. I think it was something like 60/30/10 Ph.D/MS/BS.
2. Initial simple technical online quiz
3. Two separate interviews (one case study) and one technical (stats/coding) where each was 30 minutes long.
All the questions were pretty standard though not necessarily indicative of data science performance.
Interview questions [1]
Question 1
Case study - what features would you use to determine credit risk given transaction history from the past two years.
Explain a simple map reduce problem
Read in a very large file of tab delimited numbers using python and count frequency of each number (don’t overthink this. Use python done in a few lines)
Whats more likely: Getting at least one six in 6 rolls, at least two sixes in 12 rolls or at least 100 sixes in 600 rolls?
Find all the combinations of change you can for a given amount
I applied through a recruiter. The process took 2 weeks. I interviewed at Capital One in Jul 2015
Interview
I was approached by a corporate recruiter by phone. The interview process consisted of an initial phone call with recruiter. The recruiter was pleasant and explained the job responsibilities a bit and asked if I had preference in location. The interview consists of 1) online coding challenge, 2) phone interview with a data scientist, 3) a data challenge problem, 4) onsite interview at Capital One's office. The technical interview was easy and uses Hackerrank. The phone interview was kind of confusing to me the data scientist didn't seem to have read my resume at all and didn't know I was applying to a different office. I was asked questions about basic regression models, map reduce and some data quality, biasing and munging questions. Overall I thought I did good.
Interview questions [1]
Question 1
String manipulation, permutations and dynamic programming on coding test. Phone interview was shaky as the interview didn't seem to have even read my file, preferences etc.