ML Compiler Engineer

Join a stealth-stage AI hardware startup building a custom SoC and the full compiler and inference stack alongside it. What You’ll Do You’ll own the full stack from PyTorch and Triton down to efficient machine IR on leading-edge GPUs and novel accelerator targets You’ll build and optimise ML compiler infrastructure across every layer, not just one slice You’ll drive GPU optimisation across CUDA and ROCm with real, measurable performance outcomes You’ll collaborate directly with hardware engineers on ISA and microarchitecture tradeoffs that shape the silicon itself You’ll tackle compiler problems on XPU and novel accelerator targets where no established playbook exists What We’re Looking For 3+ years in compilers, ML systems, or equivalent depth Strong compiler fundamentals with a track record of delivered performance improvements Hands‑on experience with PyTorch, Triton, and low‑level code generation GPU optimisation depth in CUDA and/or ROCm HW‑SW co‑design experience or strong evidence of meaningful hardware collaboration #J-18808-Ljbffr