
Stitch Fix Data Scientist Interview Questions
& Process
Real candidates share what happened, how many rounds they had,
and how the experience turned out.
Based on 18 interview experiences · FREE TO READ
Candidate interview experiences
First-hand accounts from people who interviewed at Stitch Fix.
Data Scientist Intern
There were 4 rounds of interviews. The first was a general phone interview on Zoom. The second round focused on statistics and ML, testing knowledge of probabilities, probability distributions, and various ML techniques. The third round was a coding challenge involving a medium-difficulty LeetCode problem. The final round was with senior team members to assess logical thinking and problem-solving abilities. The interviewers were great.
- If you were to test an offer by emailing it to 5% of your customers, how would you interpret their responses? Specifically, if 50% of that 5% accepted the offer and 50% did not, can you conclude that 50% of your entire customer base would accept the offer?
Data Scientist
I had a phone screen with the recruiter first. Then I had a one-hour interview that included both coding and technical questions. The coding question was the same one other candidates mentioned. The technical questions were about my previous projects, asking what the project was, how I deployed it, and how I evaluated it. No machine learning questions were asked. After that, there were 6 rounds of onsite interviews, which included 2 coding rounds and 4 technical rounds.
- How did you evaluate models and can you describe a past project?
- Can you tell me about a past project and how you evaluated models?
Data Scientist
We started with a chill coffee chat. Then I had two phone interviews focusing on statistical modeling. After that, I went for a full day onsite. Since I was considered for two different positions, they scheduled me for 7 hours of interviews. About half of them were about statistics, a few were about engineering, and some were about product. The stats and modeling questions were pretty good, but tough. The engineering questions were super easy.
- Could you compare the performance of two different items and recommend which one to choose for our customers?
Stitch Fix Data Scientist Interview Questions
Quoted word for word from Stitch Fix interview reports.
“What are the effects of freezing layers when fine-tuning?”
Read reports →“Suppose we have item purchase data and user purchase probabilities are uniform. If an item was recommended 10 times and purchased 10 times, what's its long-term purchase probability?”
Read reports →“Could you write a function to sample from a multimodal distribution? You'll receive keys, weights, and the number of samples (N) to draw, and you need to return a list of N keys.”
Read reports →“Could you build a classification model using a pandas dataset to predict specific classes?”
Read reports →“Can you implement a function to sample from a multinomial distribution?”
Read report →“What statistical methods can you use to determine if the means of two groups are different?”
Read report →“Can you explain what transformer networks are?”
Read report →“Solve this easy-level coding problem.”
Read report →“Could you write code to perform a Monte Carlo simulation?”
Read report →Formats, difficulty and experience
Across all 18 Stitch Fix interview reports.