- 90 minutes technical challenge (their platform collapsed at the end of that session and I couldn't present my work) -some basic and general statistics and ML questions like the difference between random forests and gradient boosted trees.
Staff Data Scientist Interview Questions
112 staff data scientist interview questions shared by candidates
Describe how K-means works.
They asked previous projects and a Leetcode medium-level question
1. Life journey was all about the career stories, choices etc. 2. First pair programming was given a sample dataset how'd you calculate a certain metric 3. Problem solving is all about what metrics you would look at for a Shopify product 4. Second pair programming was on implementing some simpler ML model and productionizing it 5. Techincal deep dive was bascially choosing a project and talking in depth with the interviewers
You can make PPT for leadership?
What is most important for data science?
Model training process. Data leakage. Variable selection steps.
What would you do if you have unlimited resources at the company?
presentation of a Kaggel problem
Interview 1 - Rotate array in such a way that sum of index*value is max - In a connected graph, randomly clicked any button with prob p, then nearby switches state changes. What is the expected no of switches open. Interview 2 - Two fair dice with 4 faces, you win if blue > black , end if blue = black and return is max (blue-black, 0). What is the expected win amount, build some recursive solution with E = prob3*(amount + E) + prob2*(amount + E) + … . solve it via coding simulation as well. (n no of loops each can have infinite no of trials) - 2sum based, with d as the difference between the two, but avoid repetitions and count (a,b) , (b,a) as one. Interview 3 - Hotel cancellation SD, with immediate profit improvement , - [1,0,0,0,1,1,0] → no of continuous subsets 4, if by prob p. with which we can get 1, what is the expected no of subsets. Sum from 1 to n P(x)*x or use recursive formula E(n,p) = prob getting same in boundary * E(n-1,p) + prob getting diff in boundary * (E(n-1,p) + 1)
Viewing 11 - 20 interview questions