Meta 2025 Interview Research

Questions

• How much possibility for this position to convert to full-time?

• Does Meta use AI models internally beyond LLaMA, or is it limited to that?
• claude
• gpt
• flexible

• Will this role involve working directly with Meta’s internal infra such as HHVM or specific agent frameworks?

• Which benchmarks or evaluation suites are currently most important for your team, and would my work focus only on benchmark projects or on broader team initiatives as well?

Stacks

• 
  PHP

• HHVM

• Hack Lang

Projects

• MLEBench (Kaggle end-to-end)
• feature engineering
• relatively small dataset size

• Berkeley Function Calling Leaderboard (BFCL)

• I have also similar experience with agent trace to agent graph dataset with 30 data points at Holistic AI

Relatives

• BELA

  facebookresearch • Updated 2026 Feb 2 11:21

• ‣