Senior GPU Compiler Engineer (MLIR/LLVM)

SF Tensor; San Francisco (CA); On-site; USD 285,000 - 315,000; Full time

Benefits offered by this job

Relocation assistance

Job summary

SF Tensor is building the ultimate GPU compiler and model factory. We hire researchers and engineers who want to drive end-to-end GPU compilation, MLIR dialect development, and low-level optimizations across NVIDIA, AMD and TPU targets.

Join a tight-knit team working on a world-class stack—StableHLO ingestion, MLIR passes, backend code generation, and direct cubin emission—while benefiting from relocation assistance and a path to-scale impact on AI compute performance.

Qualifications

• Deep experience with compiler infrastructure (LLVM/MLIR or similar).
• Strong background in GPU architecture and low-level optimization (CUDA, ROCm or similar).
• Hands-on experience with at least one GPU ISA (PTX/SASS, GCN/RDNA) is preferred.

Responsibilities

• Build and extend MLIR dialects and passes to optimize training and inference workloads.
• Work in the LLVM backend on instruction selection, scheduling, register allocation and cubin emission; support multiple targets (AMD, TPU, Trainium).
• Expand search-based compiler infrastructure including agent- and RL-driven program search and formal correctness proofs.
• Implement classic compiler optimizations for large-scale training.
• Create hybrid codegen paths where direct MLIR lowering isn’t practical.
• Own testing, benchmarking and performance regression systems; ensure bit-identical hardware models.
• Collaborate closely with research teams and customer workloads to identify optimization opportunities.

Skills

Compiler infra

GPU architecture

GPU ISAs

ML compiler stacks

C++/Rust

Production-grade infra

Tools

LLVM

MLIR

CUDA

ROCm

PTX/SASS

Job description

SF Tensor is building the ultimate GPU compiler and model factory. We hire researchers and engineers who want to drive end-to-end GPU compilation, MLIR dialect development, and low-level optimizations across NVIDIA, AMD and TPU targets.

Join a tight-knit team working on a world-class stack—StableHLO ingestion, MLIR passes, backend code generation, and direct cubin emission—while benefiting from relocation assistance and a path to-scale impact on AI compute performance.