Skip to main content
T0 Big tech ByteDance Rounds 1-2

ByteDance AML AI Infra rounds 1–2

Real ByteDance AI Infra rounds 1–2 interview questions covering AI Infra

Round 1

Inference optimization

  1. What are the main performance bottlenecks in large-model inference?
  2. What core optimization techniques exist for large-model inference?
  3. How does PagedAttention work?
  4. What is the design idea behind Orca’s iteration-level request scheduling?

Project experience

  1. Introduce your project experience (MPS, training memory management, scheduling strategies, time-division multiplexing and space-division multiplexing, etc.).

Coding

  1. Reverse a linked list.

Round 2

Fundamentals

  1. What error does an out-of-bounds array index cause in C++?
  2. On Linux, how do you debug programs and locate errors?

Project experience

  1. Introduce your project experience (MPS, GPU utilization metrics, etc.).

Coding

  1. Implement an LRU Cache.