Canva Machine Learning Engineer Interview Questions
& Process

The interview started with an HR screen where they inquired about my experience, specifically focusing on whether I had experience with TB or PB data processing pipelines. This was followed by an AI programming interview that involved removing gender bias from recommendations. After that, there were four additional one-hour interviews: one for ML System Design, one for ML System review, a pair programming session without AI, and a behavioural interview. I also had a discussion with an Engineering Manager about the role, which they characterized as primarily a software engineering position with no model training involved. Each interview was one-on-one. In one instance, the interviewer's poor communication skills caused confusion and wasted time. This also meant the outcome might have been subject to bias due to the lack of interviewer alignment. A major disappointment was the absence of constructive feedback.

I had a disappointing interaction with one of the recruiters. When I mentioned I use VS Code as my code editor, the HR person smirked, which felt condescending. They also asked, “Do you even have a job now?” — which was unnecessarily rude and unprofessional. This made me feel like not all HR team members uphold the respectful and inclusive culture Canva is known for.

The interview process involved an initial HR screening where my experience was reviewed and some technology familiarity questions were asked. This was followed by a coding interview with a senior ML engineer. The coding challenge was different from typical problems where you write code from scratch. Instead, it required working with existing code and completing specific parts. While this was a novel experience, it was challenging because I was expecting to code from the ground up. It would have been better to focus on understanding the provided code first. The interviewer wasn't very helpful in guiding me to focus on the existing code rather than trying to write new code.