Graph Optimization Engineer

Compiler Optimization Engineer / Well-funded start-up / Greenfield opportunityExciting chance to join a well-funded startup building a hardware-agnostic AI compiler allowing you to write once, run on any accelerator architecture at full performance.We're hiring a core engineer to own the graph optimisation layer of our compiler stack. You'll have real ownership over how the next generation of AI models perform across diverse and emerging hardware targets.About the role:You'll design and implement graph-level transformation passes — operator fusion, layout propagation, tiling, dead code elimination, and constant foldingYou'll get the chance to define and evolve our intermediate representation (IR) as ML model architectures and hardware capabilities advanceYou'll dig into real performance data to surface optimisation gaps and deliver measurable gains in throughput and latencyYou'll get the chance to build out testing and validation infrastructure to guarantee correctness across optimisation passes and hardware targetsYou'll work closely with frontend and code generation teams to keep IR interfaces clean and pipelines well-structuredYou'll get the chance to prototype and propose new optimisation strategies as the model and hardware landscape evolvesKey Requirements:You'll hold a degree in CS or Computer Engineering (BS, MS, or PhD)You'll have strong hands-on C/C++ experience in performance-critical environmentsYou'll bring deep knowledge of graph-level compiler optimisation — fusion, tiling, layout transformations, DCEYou'll be able to speak to concrete, measurable performance outcomes from your workIt's a big plus if:You have hands-on experience with MLIR, XLA, or comparable graph-level IR frameworksYou're familiar with ML framework internals — PyTorch eager/compile mode, JAX/XLA, or TensorRTYou've explored polyhedral models or affine analysis for loop and tensor optimisationYou understand hardware memory hierarchies and how layout choices drive performance on GPUs and acceleratorsYou've worked with quantisation, sparsity, or other model-level optimisation techniquesYou've contributed to open-source compiler or ML infrastructure projects