NVIDIA

Deep Learning Compiler Engineer - CUDA

NVIDIA

Beijing, China · Full Time

Be the first to apply

Experience
2+ yrs
Salary
Openings
1
Posted
4 weeks ago
Work mode
In office
Education
Masters or PhD
Resume
Required to apply

Where you'll work

Sign in to tell us what does and doesn't work for you here — it sharpens every match we show you.

Job description

About the Company

NVIDIA has been at the forefront of computer graphics, PC gaming, and accelerated computing for over 25 years, pioneering innovations fueled by advanced technology and talented individuals. Currently, NVIDIA is channeling AI's vast potential to shape the future of computing, enabling GPUs to serve as the intelligence powering computers, robots, and autonomous vehicles capable of understanding their environments. Joining NVIDIA means becoming part of a diverse and supportive environment dedicated to excellence and impactful work.

Job Overview

We are seeking a cuTile Core Compiler Architect to join our NVIDIA Architecture team. This group is focused on developing cutting-edge architectures and infrastructure to maintain NVIDIA's leadership in graphics and parallel computing by driving innovation in parallel programming models and GPU architectures.

Key Responsibilities

  • Architect and implement the domain-specific language (DSL) and core compiler for tile-aware GPU programming models tailored to upcoming GPU architectures.
  • Continuously refine and enhance the compiler’s core architecture to improve overall performance.
  • Research and propose solutions for next-generation GPU architectures within the DSL and compiler toolchain.
  • Conduct performance evaluations on emerging AI and large language model workloads, integrating findings with AI/ML frameworks.

Required Qualifications

  • Master’s or PhD degree (or equivalent experience) in Computer Engineering, Computer Science & Engineering, Computer Science, Artificial Intelligence, or related disciplines.
  • At least two years of pertinent professional experience.
  • Proficient in C and C++ programming with strong software engineering capabilities; background in ACM is advantageous.
  • Solid foundational knowledge of computer architecture principles.
  • Excellent analytical skills with the ability to abstract and solve complex problems effectively.
  • Expertise or strong experience in compiler technologies such as MLIR, TVM, Triton, or LLVM is highly desirable.
  • Familiarity with GPU architecture and the ability to perform high-speed kernel programming is beneficial.
  • Understanding of large language model algorithms or expertise in specific high-performance computing domains is a plus.
  • Knowledge of multi-GPU distributed communication frameworks adds value.
  • Strong verbal communication skills in English are preferred.

Additional Information

NVIDIA is recognized globally as a highly sought-after employer in technology, offering competitive salaries and comprehensive benefits packages designed to support employees and their families.

Minimum education

Doctorate

Languages

English
🤖
Online · instant AI help
Broxer