Compiler Backend Engineer

The role

You will design and build custom compilation that maps LLM inference onto hardware. This involves scheduling, memory planning, and code generation for non-CPU/GPU hardware.

What you will do

• Build passes that turn lowered programs into hardware execution code
• Own how work and data are scheduled and placed across hardware resources
• Own fusion and dispatch decisions for LLM inference
• Work the abstraction layer between the compiler and the hardware
• Validate optimizations with results from simulators, FPGA prototypes, and silicon
• Maintain a performance regression suite
• Decide which operations the compiler generates and which use hand-tuned kernels

Required qualifications

• Deep experience with compiler optimization or code generation for accelerators or other non-CPU targets
• Comfort working close to hardware
• Willingness to design compilation approaches from first principles
• Track record of shipping optimizations that moved a measured performance number
• Comfort building software for hardware that is still evolving
• Fluency in Python
• Familiarity with C, C++, or Rust

Nice to have

• Experience with PyTorch's compilation stack (TorchDynamo, FX, Inductor, torch.export) or other ML compilers, such as IREE, TVM, or XLA
• Experience compiling for in-memory, spatial, or other non-von-Neumann compute targets
• Background in hardware design tools such as logic synthesis or place-and-route
• Experience building or using cycle-level or performance simulators to guide optimization decisions
• Contributions to open-source compiler or kernel infrastructure (e.g., PyTorch, Triton, TVM, MLIR, LLVM)
• Publishing or presenting compiler or performance work at venues such as MLSys, ASPLOS, or CGO

About the company

Normal is an applied AI company solving the hardest problems in AI and silicon. They build foundational hardware and software for the semiconductor industry, critical AI infrastructure, and the broader systems that power our world.