Skip to main content
T5 Others iFlytek New grad

iFLYTEK AI Infra campus hiring

Real iFLYTEK AI Infra campus hiring interview questions covering high-performance computing and related topics

High-performance computing

Fundamentals

  1. What are FlashAttention’s core optimizations? How do tiled QKV loads, Online Softmax, and reducing memory complexity from O(N^2) to O(N) each work?
  2. In Self-Attention, why divide the dot product by sqrt(d)? What goes wrong without scaling?
  3. How are callback functions implemented?
  4. Common ways to debug out-of-bounds GPU memory access?

Project experience

  1. Deep dive into project experience.