Interview Guide for Machine Learning Engineers

This guide has been created to help you prepare for your upcoming interviews – we want to be as transparent as possible in our hiring process as we value your time.

Group of 3 Deliveroo employees sat together with laptops

There are four stages to your machine learning engineer interview process.

You’re encouraged to ask questions throughout; this should feel like a two-way interview and we’d like to make sure you have the time to ask questions too.

The below processes apply to all levels. If there are any changes to the outlined processes below, you will be notified by your recruiter.

Understand the full interview process

Recruiter Screen

For 20-30 minutes, you’ll speak with one of our tech recruiters who will walk you through the role of an MLE, the various ML teams and information about the wider business. We will also ask you about your career and motivations in your next role.

Technical Screen

For 60 minutes you will speak with a Machine Learning Engineering team member. The majority of this interview is focused on your general career trajectory so far as well as which machine learning projects you’ve worked on.

There is nothing to prepare, but you will be asked both business and technical questions on at least one specific ML project you’ve recently worked on. For example, you may be asked to talk about how you measured success in your project as well as the theory behind your modelling choices.

You will also be asked some general Python and Engineering questions towards the end of the interview.

Take-home Assignment

Your take-home assignment is a timed 2h30 test. You’ll be tasked with writing a project plan document, writing code as if you’re contributing to a production codebase, and peer-reviewing code. Once you’ve returned this, a team member will mark it and your recruiter can provide you with specific feedback around the work. If you are successful, you will be invited to complete our final round of interviews.

Final Interviews

In general, throughout each of these interviews, attempt to display energy, enthusiasm and clear & concise explanations. Do not feel as though you need to rush – begin by asking clarifying questions to ensure you know what the interviewers are expecting and then structure your solution/description.

Machine Learning Theory (1 hour)

• We will normally begin this interview by asking you to describe a previous project you have been majorly involved with which has involved machine learning. It’s likely you’ll be asked to explain your decision-making process behind choosing certain models.

• We will then deep dive into the theory behind common machine learning algorithms. We won’t be testing for expert knowledge of the most cutting-edge advances in ML, rather a solid understanding of the absolute fundamentals.
You may be asked to derive formulae in this interview and you can use a (virtual) whiteboard to communicate your understanding if desired (we use LucidSpark for this).

Machine Learning Practical (1 hour)

• In the ML practical, you will face a Deliveroo-specific problem that we have faced in ML Engineering. This should be a problem you’ve not faced before in your career, so we can see how you tackle an unstructured problem. We are hoping to test your ability to reformulate a business problem so it can be tackled using data science techniques and then discuss how you would build out an appropriate real-world solution. This is a two-way conversational interview and will cover the topics that reflect a real-world project, such as problem framing, model evaluation, model optimisation and productionisation.

Behavioural (45 mins)

• In this interview, you will face competency-style questions on how you’ve dealt with challenges and learnt from mistakes in your career so far. These are typically “Tell me about a time” questions that will encourage you to think about when you’ve dealt with challenging scenarios, such as tricky stakeholders, conflicting opinions or tight timelines. We’d advise you to follow the STAR methodology to help provide structured answers.

Additional Tips

Listen Carefully

Rephrasing questions or asking for clarity is okay, as is telling the interviewer you want time to collect your thoughts.

Be Concise

Make sure you’re answering the question and not using a prepared example that isn’t applicable or related to the questions. (the STAR method tends to work here nicely)

Thoughtful Questions

Come prepared with thoughtful questions (e.g., team priorities, culture, etc.).

Feel Comfortable

Interviewing is a two-way street, so we want you to feel comfortable and understand how a role at Deliveroo could help you grow your career.

We’re always looking for talented Machine Learning Engineers who are excited to solve meaningful problems at real scale. If you’re passionate about turning data into intelligent, production-ready systems and want to see your work power a global, three-sided marketplace, we’d love to hear from you. Explore our open roles and come build the future of Deliveroo with us.

Apply Now