i went through the Canva machine learning engineer interview process in early 2026 for a senior ML role. didn't get the offer but i made it to the final round so i got a full look at the loop. writing this up because there's almost nothing specific about ML at Canva online.
the process: recruiter screen (standard, they asked about my ML background, what i'd shipped, not a quiz) ML fundamentals round (60 min, this was a mix of theory and applied. more applied than i expected) ML system design (60 min, open-ended, design a recommendation system or an image classification pipeline at Canva's scale) coding round (45 min, this was more SWE-ish than pure ML. they wanted to see actual Python, clean code, not just notebook-quality scripts) behavioral / values (45 min)
the ML fundamentals round: they asked about things like bias-variance tradeoff, when you'd prefer precision vs recall, and how you'd handle class imbalance. nothing i hadn't seen before. but they also asked about model serving, latency constraints, and how to think about a model that performs great offline but degrades in production. that last one tripped people up.
ML system design: i got the design-a-content-recommendation-system prompt. they wanted me to walk through: feature engineering, candidate generation, ranking, how i'd handle the cold start problem for new users, and how i'd measure success. the conversation went deep on offline vs online metrics and why they don't always align. this is Canva-specific because their product is very visual, so they also asked how i'd think about image embeddings as signals.
the coding round was the surprise. they gave me a problem that was basically: implement a simple k-means clustering from scratch. no sklearn. they wanted clean Python, not just pseudocode. if your coding skills have atrophied because you spend most of your time in notebooks, practice writing actual clean Python before this.
where i think i got cut: the behavioral round. i gave answers that were technically fine but probably came across as too individual-contributor-focused. Canva's values are genuinely about collaboration and impact across teams. tailor your STAR answers accordingly.
comp ranges i saw mentioned for senior ML in Sydney: AUD 200-240k total. US remote comp benchmarks higher, probably around USD 200-230k TC but i don't have a direct data point there.