AI-assisted coding interviews are becoming the norm. Companies like Canva now require candidates to use AI tools, while Stripe, Meta, and others actively encourage it. The bar has shifted — interviewers want to see how well you collaborate with AI, not just whether you can solve the problem.
We've launched a new section on AI Coding Gym to help you practice for these interviews with real questions.
What's inside
We've curated a collection of real interview questions from companies including Meta, Stripe, Coinbase, Shopify, Canva, LinkedIn, and Rippling. These aren't generic algorithm puzzles. They're problems that candidates have actually faced in recent interviews — sourced from candidate reports, interview prep communities, and public postings.
Each problem includes the full description, progressive extensions that ramp up complexity mid-interview, and details about the AI setup — whether tools like Copilot or Claude were mandatory, recommended, or optional.
Real candidate experiences
What makes this different from a typical problem bank is the candidate data. Many problems include real reports from people who've been through the interview: their experience level, the outcome, and direct quotes about what worked and what didn't.
This kind of signal is hard to find elsewhere. It helps you calibrate expectations — not just for the technical difficulty, but for the pace, the follow-ups, and how interviewers evaluate AI-assisted work.
Progressive complexity
Most of these problems aren't single-shot. Interviewers start with a core task, then layer on extensions — additional constraints, new requirements, performance considerations. Canva's Figma clone challenge, for instance, starts with a basic collaborative design tool and progressively adds real-time collaboration, undo/redo, and persistence.
This mirrors how companies actually evaluate candidates with AI: they want to see how you manage scope, how you direct AI through evolving requirements, and how you handle the point where AI-generated code stops being good enough.
Community discussions
Every problem has a discussion thread. Share your approach, read how others solved it, ask questions. Discussions are organized by type — solutions, experiences, and questions — so you can find what's most useful. Voting surfaces the best contributions.
Interviewed somewhere recently? You can submit new problems directly on the platform to help the community stay current.
Start practicing
The best preparation for an AI-assisted interview is practice under realistic conditions. That's exactly what this is for.
Browse the questions at aicodinggym.com/community.