Machine Learning Interview Questions - Dos and Don'ts
Share this blog
Practising or aspiring machine learning engineers, you might want to ace the interviews by answering all the interview questions asked by the interviewer. Never leave a corner to lose an opportunity.
I am a Lead Research Scientist at InFoCusp Innovations Pvt. Ltd and a machine learning mentor at MentorIF. With 6+ years in Infocusp , having worked on a wide array of Machine Learning (ML) and Deep Learning (DL) projects – from financial time series modelling, real time object detection, Convolutional Neural Networks (CNNs), convex optimization, mapping Electroencephalography (EEG) signals and audio, music generation using Deep Learning (DL) and neuro symbolic systems for scene understanding/ question answering. I can offer you data science advice based on what I’ve learned over the years and if you need advice on how to integrate machine learning to your projects, regardless of the domain, as well as when not to.
A Master’s degree in ICT (Gold medalist) from Dhirubhai Ambani Institute of Information and Communication Technology(DA-IICT) and Bachelor’s degree in Electronics and Communication from Nirma University.
A member of the team that won the secondary round and main round of Kaggle’s PASSNYC data science for good challenge, placing 2nd and 5th globally.
You may count on me to assist you with any part of your machine learning journey, including projects, interviews, questions, and startup requirements. Feel free to reach out to me at MentorIF for your mentoring and training requirements.
Many people have asked me for the list of interview questions and the job expectations after I conducted over 50 interviews for the position of Machine Learning Engineer. So, in order to help, I decided to write about it. This is not an exhaustive list of questions and answers to practice machine learning interview, which I will be covering in next blog. Instead, I want to provide a list of the main criteria that, in my experience, determine whether machine learning (ML) candidates are accepted or rejected.
Your Machine Learning Resume
When referencing the projects or subjects on your resume, be thorough. In order to appear impressive, I’ve seen many people include machine learning projects they’ve never worked on in their resumes. In reality, this will just lower your chances to get shortlisted in interview rounds. By mentioning machine learning (ML) or deep learning (DL) buzzwords in your resume you won’t improve your chances of getting hired if you don’t understand what they mean.
You might have mentioned Principal Component Analysis (PCA) or K-Nearest Neighbour (KNN) in your resume, but understanding the inner workings of these machine learning techniques is crucial. A glaring red flag for the interviewer is when a candidate uses tools or functions that libraries provide without even knowing what they’re useful for. You can download a winning, simple yet effective one page resume format here.
Then, following are a few must-know concepts if you are applying for a machine learning role:
- Gradient descent
- Overfitting/ underfitting
- Loss functions, basics of optimizers (the role they play in gradient descent)
- Cross validation, regularization
- Confusion matrix
- Basic probability/ statistics and some linear algebra
- Feature cleaning/ normalization/ selection
- One or more basic Machine Learning or Deep Learning methods (whichever you’ve worked on)
And if you’ve worked in Natural Language Processing (NLP):
- The common NLP pipeline: stemmer, lemmatizer, tokenizers, common word vectorization methods.
If your work is Deep Learning (DL) heavy:
- Batchwise processing
- Loss functions, optimizers
- Hyperparameters, what to tune
- Components of Neural Networks (NN) / Convolutional Neural Networks (CNN) / Recurrent Neural Networks (RNN) (whichever you’ve used)
Simply knowing these is not enough. But your prospects of passing the interview rounds are quite slim if you don’t even understand these fundamental concepts.
Interview Questions and Answers Tips
Tip 1: Do Not Bluff
It is completely okay not knowing every concept. It’s not necessary for you to be an expert and answer all interview questions. Most of the time, interview questions are made to push your knowledge of the subject to its limit. And the interviewer might be able to tell how well you understand machine learning within the first few phrases. Long winded explanations leading to nothing will just hurt your chances of getting shortlisted.
Tip 2: Deep Learning (DL) Toolkits
Describe the tools you have used and your level of familiarity with them honestly. That sets the right expectation level for the interviewer and they know which level of interview questions to ask. Most machine learning roles require a basic level of expertise with these technologies.
You cannot have worked on machine learning or deep learning projects without coming into contact with these fundamental libraries. You haven’t worked on projects using real data if you don’t know how to use these machine learning libraries’ fundamental operations. In order to effectively respond to important interview questions, it is usually good to be familiar with the fundamental functionalities. Also, some basic understanding of data structures and algorithms is expected from you since these are ubiquitous across projects.
Tip 3: Explore Outside Of The Task Assigned to you
Knowing more about the subject, including methods you didn’t directly apply in your projects, demonstrates your interest in the subject and that you have done more research on it than is strictly necessary. For instance, in my project, I have used quicker Regions With Convolutional Neural Networks (RCNN), but I still have some knowledge about Yolo (You Only Look Once) and SSD techniques (Single-shot detector).
Tip 4: Know Some Logistics About The Datasets You Used
With the advent of deep learning, many people just use off-the-shelf machine learning (ML) methods without having explored the dataset. As a result, candidates being clueless about the scale and features of the dataset.
Bonus points if:
- You know about state of the art methods in the domain you’re working on and have explored recent trends
- Some good (even small scale) projects on GitHub/Kaggle
- You answer succinctly (Yes! very few people have the skill)
- Know linear algebra concepts
Tip 5: Some Major Red Flags (for interested readers)
- Have used Convolutional Neural Networks (CNN) for multiple tasks: Not sure why it has a dropout layer/ what it does
- Have used thesis on Generative Adversarial Networks (GANs) and unsure of what was the loss function used in them
- Bluffing about what convolution operation does in the Convolutional layer of CNN
- Have 3 years of work experience but have only used off the shelf tools over the years and don’t know how they work.
- Changing the parameter C increased the accuracy, not sure what C does
Last but not least, chances of you getting hired for a machine learning role at your dream company are very slim if company is specifically looking for candidates with more practical experience or a research background. Do your own research and ask the hiring manager these questions. If you don’t get past the interview stage, don’t let that derail your motivation. Learn from it and move forward.
There will be one more post on helpful resources to help with the above-mentioned details and machine learning interview questions. If you have any questions or recommendations, please send me a Direct Message (DM) at MentorIF and I’ll do my best to respond.
Wishing you all the very best for your journey!
P.S: Don’t google stuff during online interviews.
You can find machine learning mentors at MentorIF. These will help you ace your interviews, secure a job, projects, start-ups etc. Additionally, you can reach out to our support team. MentorIF is a great platform for Mentoring and Training. With the mentors of your choice, you can arrange for a free one-on-one mentoring session. It costs nothing to chat with mentors. Don’t think twice before sending them your first message. You can also request for training to advance your knowledge and career. You can also discuss your requirements with our support team. We will assist you by creating a training program that is specifically tailored to your needs and matching the right mentors on the training.