My 22-year old coaching client landed a new-grad Machine Learning job at a NYC startup that pays $175k per year ๐
($120k base, $40k stock, $15k annual bonus)
Here's the 17 interview questions they were asked:
(how many of these could you answer?)
($120k base, $40k stock, $15k annual bonus)
Here's the 17 interview questions they were asked:
(how many of these could you answer?)
๐๐๐๐ก๐ข๐ง๐ ๐๐๐๐ซ๐ง๐ข๐ง๐ ๐ค
1. What is PCA, why is it helpful, and how does it work?
2. What is heteroskedasticity, and how can you check for it?
3. After loading the Boston house prices dataset (it comes with scikit-learn), can you find any outliers in the data?
1. What is PCA, why is it helpful, and how does it work?
2. What is heteroskedasticity, and how can you check for it?
3. After loading the Boston house prices dataset (it comes with scikit-learn), can you find any outliers in the data?
๐๐๐๐ก๐ข๐ง๐ ๐๐๐๐ซ๐ง๐ข๐ง๐ (continued) ๐ค
3 (cont). What should we do with these outliers & why?
4. On the Boston house prices dataset, can you build a simple regression model, and interpret the model's performance?
3 (cont). What should we do with these outliers & why?
4. On the Boston house prices dataset, can you build a simple regression model, and interpret the model's performance?
๐๐ ๐๐ฒ๐ฌ๐ญ๐๐ฆ ๐๐๐ฌ๐ข๐ ๐ง โ๏ธ
5. Imagine your building a system to recommend users items similar to the items theyโve bought before. How would you go about building this?
6. For that recommender, how would you handle a new user who hasnโt made any past purchases?
5. Imagine your building a system to recommend users items similar to the items theyโve bought before. How would you go about building this?
6. For that recommender, how would you handle a new user who hasnโt made any past purchases?
๐๐ ๐๐ฒ๐ฌ๐ญ๐๐ฆ ๐๐๐ฌ๐ข๐ ๐ง (continued) โ๏ธ
7. Let's say you had to scale this recommender to 1 million users, the product catalog had 100k items, & 99% of users had bought less than 3 items before.
How would this change your answer?
7. Let's say you had to scale this recommender to 1 million users, the product catalog had 100k items, & 99% of users had bought less than 3 items before.
How would this change your answer?
๐๐ ๐๐ฒ๐ฌ๐ญ๐๐ฆ ๐๐๐ฌ๐ข๐ ๐ง (continued) โ๏ธ
8. What are feature and concept drift?
Give me an example related to the product recommendation system earlier.
How would prevent this drift issue?
8. What are feature and concept drift?
Give me an example related to the product recommendation system earlier.
How would prevent this drift issue?
๐๐ฒ๐ญ๐ก๐จ๐ง ๐
9. Given a list of integers called nums, return all the triplets [nums[i], nums[j], nums[k]] such that i ! = j, i != k, and j != k, and nums[i] + nums[j] + nums[k] == 0.
Example Input: nums = [-1,0,1,2,-1,-4]
Example Output: [[-1,-1,2],[-1,0,1]]
9. Given a list of integers called nums, return all the triplets [nums[i], nums[j], nums[k]] such that i ! = j, i != k, and j != k, and nums[i] + nums[j] + nums[k] == 0.
Example Input: nums = [-1,0,1,2,-1,-4]
Example Output: [[-1,-1,2],[-1,0,1]]
๐๐ฒ๐ญ๐ก๐จ๐ง (cont.) ๐
10. Given a string s, find the length of the longest substring without repeating characters.
Input: s = "abcabcbb"
Output: 3
(the answer is 3 because the longest string is "abc")
10. Given a string s, find the length of the longest substring without repeating characters.
Input: s = "abcabcbb"
Output: 3
(the answer is 3 because the longest string is "abc")
๐๐ฒ๐ญ๐ก๐จ๐ง (cont.) ๐
11. Given a Pandas data-frame of integers, with m rows and n columns, if an element in the data-frame is 0, set its entire row and column to 0's.
Do it in place.
11. Given a Pandas data-frame of integers, with m rows and n columns, if an element in the data-frame is 0, set its entire row and column to 0's.
Do it in place.
๐๐ข๐ฌ๐๐๐ฅ๐ฅ๐๐ง๐๐จ๐ฎ๐ฌ ๐คทโโ๏ธ
12. I see you listed Kubernettes on your resume.
What did you use it for & how was your experience?
What frustrated you?
13. I see you listed TensorFlow on your resume.
What do you like about it?
What would you change about TensorFlow?
12. I see you listed Kubernettes on your resume.
What did you use it for & how was your experience?
What frustrated you?
13. I see you listed TensorFlow on your resume.
What do you like about it?
What would you change about TensorFlow?
14. Walk us through your past internship project?
What was the most challenging part of getting your model deployed?
15. Walk us through your undergrad research project.
Why did you pick XGBoost?
(cc: @tunguz)
What other models did you try?
What was the most challenging part of getting your model deployed?
15. Walk us through your undergrad research project.
Why did you pick XGBoost?
(cc: @tunguz)
What other models did you try?
@tunguz 16. I saw you took Linear Algebra in college.
What is an eigen-value? What about an eigen-vector?
Can you give me an example of an ML technique that makes use of these concepts?
17. Why do you want to work at this company?
Why a startup? Why not grad school?
What is an eigen-value? What about an eigen-vector?
Can you give me an example of an ML technique that makes use of these concepts?
17. Why do you want to work at this company?
Why a startup? Why not grad school?
@tunguz That's a wrap!
For more Data Science & ML resources:
1. Follow me @NickSinghTech
2. Join my 9-day Data Interview Crash Course: bit.ly
3. RT the tweet below to share these questions with your followers!
For more Data Science & ML resources:
1. Follow me @NickSinghTech
2. Join my 9-day Data Interview Crash Course: bit.ly
3. RT the tweet below to share these questions with your followers!
Loading suggestions...