Meta · Phone screen
In processMeta PhD SWE Intern AI Coding Phone Screen: 1 Experience
A late-2026 Meta PhD software engineer intern phone screen: one AI-enabled coding round on friend recommendation, with two AI models to pick from.
At a glance
- Process: a single technical phone screen, run as an AI-enabled coding round; the candidate applied online and passed it (report 01)
- Asked recently: a friend recommendation task (report 01)
- Difficulty: the candidate rated the screen average
- Practise first: Mutual-Friend Recommendations
Coverage: One PhD student about to graduate who applied online for a Meta software engineer internship. The report covers one technical phone screen in late 2026 and went up on November 21, 2026. The candidate passed.
What came up
The whole screen was one AI-enabled coding round built around friend recommendation (1 of 1 reports, Nov 2026 (report date)).
Questions we found
AI Coding
| Question | Reports | Last seen |
|---|---|---|
| Friend recommendation (practice) | 1 | Nov 2026 (report date) |
1 detailed candidate timeline, loops, and screens
01: PhD SWE intern, AI-enabled coding phone screen
Interview period: 2026 Q4 · Report date: November 21, 2026 · Role: Software engineer intern · Degree listed: PhD · Channel: Applied online · Candidate's rating: Positive experience, average · Result: Passed
AI-enabled coding round: friend recommendation
- The phone screen was one coding round with AI help allowed, and the task was friend recommendation. The report gives only the task's name, not its rules or follow-ups.
- FastPrep's closest practice page is Mutual-Friend Recommendations, an earlier Meta friend recommendation task: from a list of friend pairs, suggest people who are not yet the user's friends and share at least one mutual friend with them, ranked by how many mutual friends they share, with ties broken by the smaller user ID, and return the top k. It is a plain coding problem with no AI tools, and the report does not say whether this candidate's version used the same rules.
- Two AI models were on offer in the round: Claude Opus 5 and GPT Astra. In the candidate's experience, Opus tended to add edits and extra code nobody asked for, while Astra worked a little better under interview conditions.
Result
- The candidate passed the phone screen.
Common questions
Can you use AI in the Meta intern phone screen?
Yes, in report 01. The screen was an AI-enabled coding round with Claude Opus 5 and GPT Astra available, and the candidate preferred Astra.
What coding task does Meta give in the PhD intern phone screen?
Report 01 got friend recommendation; the report names the task but not its rules.
What to practise
- Mutual-Friend Recommendations. The closest practice version of the friend recommendation task in report 01, November 2026 (report date).
Keep practicing
- Practice question: Mutual-Friend Recommendations (report 01).
- More Meta coding questions: the Meta coding question bank. It needs sign-in.
- Meta AI coding projects: the Meta project coding bank. It needs sign-in.
- More on this round type: the Meta AI-Native Coding Interview article, from full-time candidates.
- Meta full-time report: Meta Business Support Engineer Phone Screen: 1 Experience, a separate article for experienced full-time candidates.
- Interview overview: the Meta interview questions page.
- Hiring insights: the Meta hiring insights page.
- More firsthand reports: the Experience feed.
Sources: Anywhere useful interview info can be found.
Comments
0No comments yet — be the first to share what you know.