Compiler Backend Engineer
Core
Design and build custom compiler backends to map LLM inference onto novel in-memory hardware, optimizing scheduling, memory planning, and code generation.
Role type
Senior IC compiler backend engineer (accelerator/hardware)
Builds
Custom compilation pipelines for LLM inference on non-von-Neumann hardware
Domain
AI infrastructure / Semiconductor / Compiler technology
Deliverable
production ML models
Required skills
Compiler optimization, code generation, scheduling, memory planning, fusion, dispatch, hardware co-design, profiling, validation, Python, C/C++/Rust
Preferred skills
PyTorch compilation stack (TorchDynamo, FX, Inductor), ML compilers (IREE, TVM, XLA), in-memory compute targets, hardware design tools, cycle-level simulators, open-source contributions
Technologies
Python, C, C++, Rust, PyTorch, TorchDynamo, FX, Inductor, torch.export, IREE, TVM, XLA, MLIR, LLVM
Responsibilities
Build passes to generate efficient code for hardware operations; Own scheduling and memory placement to minimize latency; Make fusion and dispatch decisions to maximize compute utilization; Collaborate with hardware architects on abstraction layers; Validate optimizations using simulators, FPGAs, and silicon; Decide between compiler-generated operations and hand-tuned kernels
Seniority
Senior, hands-on IC
