Skip to main content
T4 Automotive and self-driving Li Auto Internship Round 1

Li Auto Cloud AI Infra internship first round

Real Li Auto AI Infra internship first-round interview questions covering inference and operator optimization

Inference optimization Operator optimization

Round 1

Fundamentals

  1. What is the principle and role of Batch Normalization?
  2. What causes vanishing and exploding gradients, and how do you address them?
  3. What optimization methods exist at the low-level operator layer?

Project experience

  1. Introduce the project; explain what you did for training optimization and inference optimization.

Coding

  1. Given array nums and target, find two numbers that sum to the target.

Round 2

  1. What mainstream methods exist in object detection?
  2. What later paper improvements exist for LSS? At deployment time, how do LSS and BEVFormer compare in compute cost?
  3. How do Diffusion-based BEV methods differ from BEVFormer? (BEV evolution: gen-1 LSS -> gen-2 BEVFormer -> gen-3 Diffusion-based)
  4. Team directions: training acceleration and optimization, inference optimization, forward-looking model exploration (compute exploration).

Round 3

  1. Introduce school projects, research experience, and internship work.
  2. If you owned inference optimization now, how would you approach it? What related experience do you have?