Long process and overly rigid assessment. It felt like interviewers had specific answers in mind, especially in system design, and marked different but valid approaches negatively. Clarifying requirements or using different terminology also seemed to count against candidates, which made the experience frustrating.
Interview questions [1]
Question 1
The process also included theory-heavy questions such as deriving equations, local minima, and L1/L2 regularisation.
I applied through a recruiter. The process took 4 weeks. I interviewed at Deliveroo (Hyderābād) in June 2024
Interview
The interview process is different from the typical processes where the judgement is based on solving a hard problem. Here the assessments are based on practical knowledge and skills. There were 5 rounds: screening, assessment, ML theory, ML system design, and behavioural.
Interview questions [4]
Question 1
Screening: Questions about current projects, python, neural networks.
ML Theory: Tested theoretical knowledge and core basics of Machine Learning. Topics on logistic regression, different types of loss functions, bias, variance, neural networks, regularisation and its usage.
I applied through a recruiter. The process took 1 week. I interviewed at Deliveroo (London, England) in Oct 2022
Interview
It was meant to be in three stages but I didnt pass the first stage as the interviewer felt he didnt get enough signal during the interview. The first stage used to be just project discussion but python and engineering questions have now been added which was not communicated to me prior to the interview. I mentioned that to the interviewer who apologised but still evaluates my responses.
Interview questions [12]
Question 1
In three tiers:
- Project questions:
Project questions (Based on your project)
What algorithm could be used to fit data to memory when training logistic regression?