LLVM Compiler Engineer for ML Accelerators

Meta Careers; Bellevue, Northern (WA, KY); Hybrid; USD 140,000 - 200,000; Full time

Job summary

Meta Platforms seeks a Software Engineer to join the MTIA LLVM Compiler team in Bellevue, WA, building a production compiler stack for custom AI accelerators. You will shape compiler passes, transform LLVM IR, and ensure correct, high-performance code across PyTorch models to optimized binaries for hardware backends.

You will work with compiler, ML, firmware, and ASIC experts to balance compile time, runtime performance, and code size while iterating on validation and benchmarking across

Qualifications

• Experience designing and implementing compiler passes.
• Ability to map ML workloads to hardware backends.
• Strong debugging and performance profiling skills.

Responsibilities

• Design and implement LLVM compiler passes for ML accelerator backends.
• Analyze and improve code quality with profiling and benchmarking.
• Develop LLVM IR transformations for efficient mapping to hardware.
• Collaborate with hardware and ML framework teams to ensure correctness.

Skills

LLVM compiler

Optimization

Code generation

IR transformations

Tools

LLVM

Profiling tools

Job description

Meta Platforms seeks a Software Engineer to join the MTIA LLVM Compiler team in Bellevue, WA, building a production compiler stack for custom AI accelerators. You will shape compiler passes, transform LLVM IR, and ensure correct, high-performance code across PyTorch models to optimized binaries for hardware backends.

You will work with compiler, ML, firmware, and ASIC experts to balance compile time, runtime performance, and code size while iterating on validation and benchmarking across