Candidates keep reporting on their AI-assisted rounds, and they don’t all look the same.

Take one FAANG+ Company’s OA. A candidate who got through it said there were two questions. One was an easy LC question, and the other was an AI-assisted question that was a medium DP problem. The AI helped with syntax and simplifying the problem statement.

Then there's that company’s AI Coding Rounds. A candidate described their SWE intern coding round as an AI round where you're allowed to use your own AI tools, and they named Codex, Claude, and Cursor. You get a problem statement, and it's your job to make it work with your coding tool. They compared it to system design, since you also have to think about how your code scales, what's missing, and what it would take to run in production.

Figma has one for their final round. The advice we've passed along: don't try to become a React expert, and practice tracing one user action all the way through the app. What helped other candidates most was showing that you're the one driving the AI. Give it small, bounded tasks, inspect every change, test everything, and correct it when it's wrong. Keep talking even while the agent is working. 

We're also hearing that AI-assisted rounds show up in both OAs and final rounds at a lot of companies now. This is candidate intel… not an announcement from the companies. A candidate also reached out about an AI-assisted coding interview coming up for Millennium's AI engineering internship, but we don't have anything to confirm it yet.

Even though not every company has one (and the format varies a lot), it’s still worth putting time into preparing for the possibility of an AI-assisted round. Several companies have updated their process to include them, which means they’re not going away anytime soon.

According to community reports, Meta’s interviews might be AI-enabled (you're given a problem and can use AI to solve it), and the OA will still be a LeetCode-style problem. Meta is reportedly looking at how you structure your prompt and how you evaluate the AI's output. The same message also says Meta is no longer doing referrals for internships and new grad, that attending Meta events is encouraged because your profile gets tagged, and that the full process should take 6-8 weeks, although this is currently unconfirmed.

A candidate reports getting a return offer from Amazon, and credits prepping for the Gen AI fluency round as what got them through the interview.

Datadog's new-for-2026 process is reportedly has a 90-minute session with an engineer inside a real web app codebase. You explore the codebase and explain how data moves from backend to frontend, talk through designing a new feature at a high level with no whiteboard, and review a real pull request. You can code in the language you know best, and the reports say their prep material told them to study REST API design.

DoorDash’s schedule reportedly includes a 1-hour AI Code Craft round, a 1-hour Interactive Coding Round, a 30-minute Engineering Values round, and a 30-minute break.

Topics can include strings, sorting/searching, binary search, recursion, divide and conquer, greedy, DP, BFS/DFS, arrays, linked lists, stacks/queues, maps, trees, heaps, and graphs, per one of Google's on-campus info sessions. Interviewers have access to a huge question bank and can ask you anything from it. Any lean toward graph problems is probably anecdotal, so be ready for everything. Focus on clear thinking, your approach, and time and space complexity.

If you've had an AI-assisted OA or interview this cycle, reply with the company, whether it was an OA or a later round, whether the AI was built in or your own tool, and what the problem looked like. If you know how you were evaluated, include that too. And if you've heard anything about Millennium's AI-assisted round, send that along.

Subscribe to Direct Consideration to get the urgent roles, recruiter intel, events, and opportunities worth knowing about as they happen.