Machine learning interview questions
Going down some intellectual rabbit holes yields potentially useful results. Other times the results that are yielded from that rabbit hole are highly questionable. This time around I dug into machine learning interview questions with reckless abandon. There are a few cottage industries that have sprung up related to coding interview preparation. I did not spend time looking at interview preparation services and instead looked at content generally available. For example, KDnuggets, a popular technology education website, had a post about popular machine learning interview questions [1]. Strangely enough in the years I have been studying machine learning none of those questions have ever been asked of me or by me to anybody. The other list that stood out to me was from “indeed” and it shared a more deep learning skewed set of potential interview questions [2]. Most of this content seemed to be more or less related to the transactional sharing of knowledge on the subject.
A lot of the interview process is now moving beyond the transactional exchange of academic style information and some simple question and answer into other more involved interview methods. You might see a job interview project that is assigned and brought back to the interview. That method is asynchronous. Some methods of interviewing are now moving beyond that to more real time methods of interviewing like pair programming, live programming tests, or several hours of coding as a part of a working interview. Based on the highly specialized nature of machine learning technology being customized to user pipelines, models, and setups you can imagine that having somebody demonstrate the ability to work in the space could prevent a lot of false starts in the hiring process.
One of the bright spots that I noticed during my research this week was that I got to watch a bunch of videos on YouTube about “Machine Learning Interview Questions” [3]. Some of these videos have pretty high production value and others are really just somebody talking to a camera and sharing that content online. The entire process is about people trying to demonstrate mastery of machine learning to other people. Certifications are not the coin of the realm within this process. Given the massive proliferation of published academic papers in the space, that might not be enough to demonstrate mastery either. Some of these jobs gate the candidates by requiring a doctorate or certain experience requirements.
Very savvy interviewers and recruiters might even have an expectation of contributions to open source software and check a candidate's GitHub profile commits. That type of candidate research can very quickly separate seasoned developers from newcomers to machine learning. I say that because a lot of the foundational elements happened over the last 5 years and the people that were core to that effort got a chance to contribute at a much higher rate during development. A lot of my research into MLOps has involved looking at the contributors to those major open source projects on GitHub. I have been tracking that in detail and doing analysis on it for some time. You can see the main contributors to really interesting projects like https://github.com/huggingface and https://github.com/EleutherAI. Chances are however that the candidates applying to your machine learning jobs are not going to be these highly skilled and easily verified contributors to major projects in the space.
Links and thoughts:
“Ep 48 | Fast-Track labeling with Data Labeling Services in Azure ML”
“Microsoft Acquires Activision Blizzard: What's Next?”
From Google Cloud Tech, “How to find your bugs sooner with continuous integration #Shorts”
From Yannice Kilcher, “Dynamic Inference with Neural Interpreters (w/ author interview)”
Top 5 Tweets of the week:
Footnotes:
[1] Daoud, M. (2021, January). Popular Machine Learning Interview Questions. KDnuggets . Retrieved January 21, 2022, from https://www.kdnuggets.com/2021/01/popular-machine-learning-interview-questions.html
[2] Indeed Editorial Team. (2021, December 30). 30 Deep Learning Interview Questions (and Sample Answers). indeed. Retrieved January 22, 2022, from
https://www.indeed.com/career-advice/interviewing/deep-learning-interview-questions
[3] Google. (n.d.). Machine Learning Interview Questions. YouTube Search. Retrieved January 22, 2022, from https://www.youtube.com/results?search_query=machine+learning+interview+questions
What’s next for The Lindahl Letter?
Week 54: What is a Chief AI Officer (CAIO)?
Week 55: Who is acquiring machine learning patents?
Week 56: Comparative analysis of national AI strategies
Week 57: How would I compose an ML syllabus?
Week 58: Teaching or training machine learning skills
I’ll try to keep the what’s next list forward looking with at least five weeks of posts in planning or review. If you enjoyed reading this content, then please take a moment and share it with a friend.