Skip to main content
T4 Automotive and self-driving DeepRoute.ai New grad Round 1

Nullmax AI Infra campus recruiting first round (1)

Real Nullmax AI Infra campus recruiting first-round interview questions covering operator optimization

Operator optimization

First round

Fundamentals

  1. Have you defined a custom MLIR Dialect?
  2. Did the project take design cues from torch-mlir?
  3. Did you consider handling dynamic graphs?
  4. After converting to TOSA and Tensor Dialect, what Dialects do you lower to next?
  5. How does one-shot bufferization differ from dialect-based bufferization?
  6. Do you know how LLVM’s isa and dyn_cast work?

Project experience

  1. Deep discussion of related projects and internship experience.

Coding

  1. Given a compute graph, compute the minimum memory needed to run it.

Second round

Fundamentals

  1. What is the difference between static and dynamic graphs?
  2. How does MLIR handle in-place operations?
  3. Do you have CUDA Kernel development experience?

Coding

  1. Implement topological sort.