- https://www.hackingnote.com/en/interview/system-design-interview-questions
- https://github.com/donnemartin/system-design-primer#index-of-system-design-topics
A collection of useful resources for Machine Learning System Design
-
Google Machine Learning Crash Course: https://developers.google.com/machine-learning/crash-course/production-ml-systems
-
https://github.com/chiphuyen/machine-learning-systems-design
-
https://www.ritchieng.com/machine-learning-systems-design/ (Spam classifier example, error analysis, skewed data, precision, recall and large data sets.)
-
Machine Learning System Design at Large Scale: https://www2.eecs.berkeley.edu/Pubs/TechRpts/2017/EECS-2017-219.pdf
-
A collection of useful resources from top companies such as Facebook, Twitter, Google, Airbnb, Uber, Instagram, Netflix, AWS, Spotify: https://becominghuman.ai/machine-learning-system-design-f2f4018f2f8
-
https://resources.workable.com/machine-learning-engineer-interview-questions
-
Amazon Machine Learning Engieer Role: https://towardsdatascience.com/the-amazon-machine-learning-engineer-interview-7def22233382
-
What is a recent paper you read that you are excited about?
-
Given some situation with data and labels, ask what kind of architecture they'd use to solve the problem. What is SGD in your own words? How can dropout be useful in a network? What is the use for a 1x1 convolution? --- How do you keep yourself updated?
--- What are your latest projects in ML? Pick one, pet project, interesting work stuff, what ever.
-- Do you deal with big data? if so, in what capacity?
-- What are your go to packages?
-- How do you evaluate your model? (CV, ROC curve, accuracy, sampling, etc. etc. )
-- When would you use ML? When would you use NN?
-- Tell me a big or expensive mistake you've made.
-- What techniques do you use to fine tune hyper parameters?
-- Give then an example data frame, see what the fuck they'll do with it.
Build a recommendation system that shows relevant products to users
Build a visual understanding system for a self-driving car
Build a search-ranking/ ads ranking system
Lets say I am having CTR and want to maximize the # of clicks, Qs like (1) How do we select features (2) Why do we select particular storage type for data over others (3) In terms of HP tuning, how do explore the HP space and select baseline model (4) How do we address missing values in data would be helpful.