
Blend360 Data Scientist Interview Questions
& Process
Real candidates share what happened, how many rounds they had,
and how the experience turned out.
Based on 43 interview experiences · FREE TO READ
Candidate interview experiences
First-hand accounts from people who interviewed at Blend360.
Data Scientist
The first stage involved a meeting with two individuals. It began with an introduction to the project, followed by general introductions. Then, we delved into my background. Both interviewers appeared quite disinterested and unapproachable, possibly due to a perceived lack of connection. The questions they asked felt somewhat forced, and they didn't show much interest in my responses. While it started as a discussion, it became quite uncomfortable as the interviewers didn't seem fully engaged. It felt like they were going through the motions for diversity and inclusion purposes. Towards the end, they had to rush off to another meeting. They mentioned a second stage, but the interview didn't progress that far. The interviewers gave off a somewhat racist vibe.
- Can you discuss your background?
- Can you tell me about your projects?
Data Scientist
I went through two online coding tests first. After that, I had three interviews. The first one was about basic Python and SQL. The second interview was about my past projects and I had to discuss the concepts in detail. The last interview was a behavioral one.
- Can you tell me some basic things about Python?
- How do recommendation systems work?
- What do you understand by cosine similarity?
Data Science Manager
The interview started with a recruiter call, followed by a meeting with the Director. The final round consisted of 6 parts: 5 individual interviews and a panel interview. The process felt like they were trying to extract ideas. During the presentation, they focused on getting important info about my current work. Some interviewers were rude, and surprisingly, some lacked basic knowledge of data science concepts like prompt engineering and couldn't grasp business objectives or evaluation metrics without repeated explanations.
- Given a train/test event rate of 10% and an out-of-time test data event rate of 5%, why did you proceed with model building without verifying the real-time data for the subsequent six months, which showed a 5% event rate?
- Regarding building a prospect model for targeting clients for house refinances, at which phase would you develop this model: during underwriting or funding?
- I discussed building a summarization project with Llama2 for a finance company using call transcripts. How is this feasible given that individuals might discuss SSNs and sensitive company information?
Blend360 Data Scientist Interview Questions
Quoted word for word from Blend360 interview reports.
“What are stopwords in the context of Natural Language Processing?”
Read reports →“Given a confusion matrix, how would you calculate metrics like recall?”
Read reports →“What makes logistic regression a better choice in certain scenarios compared to random forest?”
Read reports →“Can you explain the difference between logistic regression and random forest, and in which situations logistic regression might be preferable?”
Read reports →“What are the fundamentals of Machine Learning and Data Science?”
Read report →“What are the overall steps involved in building an ML model?”
Read report →“Regarding building a prospect model for targeting clients for house refinances, at which phase would you develop this model: during underwriting or funding?”
Read report →“What are the fundamentals of Database Management systems and SQL?”
Read report →“Could you explain a SQL query involving a GROUP BY clause?”
Read report →Formats, difficulty and experience
Across all 43 Blend360 interview reports.