Deep Learning Compiler Engineer - TensorIR & CUDA Tile

NVIDIA; Washington; On-site; USD 152,000 - 242,000; Full time

Benefits offered by this job

Equity

Job summary

NVIDIA in Washington state is hiring software engineers for the Tensor IR & CUDA Tile team to advance GPU compiler technology. You will work on CUDA Tile and TensorIR to generate high-performance kernels across generations of NVIDIA GPUs.

The role emphasizes compiler transformations, MLIR dialects, and performance optimization, with 3+ years in compiler work, strong C/C++ skills, and the ability to work independently in a collaborative, product-driven environment.

Qualifications

• BS/MS/PhD in Computer Science, Computer Engineering, or allied field.
• 3+ years in compiler optimization and IR design.
• Ability to work independently and lead development efforts.
• Strong C/C++ programming and debugging skills.
• Collaborative, product-focused team player.

Responsibilities

• Work on CUDA Tile and TensorIR compiler technologies for NVIDIA GPUs.
• Design and implement compiler transformations and lowering passes.
• Develop MLIR-based dialects and optimize tile-based kernels.
• Define public APIs and contribute to performance tuning.

Skills

C/C++ programming

Compiler optimization

IR design

Performance analysis

Independent work

Education

Bachelor's/Master's/PhD in CS/CE or equivalent

Tools

MLIR

LLVM

CUDA Tile

Job description

NVIDIA in Washington state is hiring software engineers for the Tensor IR & CUDA Tile team to advance GPU compiler technology. You will work on CUDA Tile and TensorIR to generate high-performance kernels across generations of NVIDIA GPUs.

The role emphasizes compiler transformations, MLIR dialects, and performance optimization, with 3+ years in compiler work, strong C/C++ skills, and the ability to work independently in a collaborative, product-driven environment.