Machine Learning Compiler Engineer
Core
Design and develop an internal ML compiler stack to transform research ideas and novel architectures into high-throughput training jobs and low-latency inference for a global trading business.
Role type
Senior IC machine learning compiler engineer
Builds
Internal ML compiler stack for training and inference
Domain
Financial technology / Machine Learning Infrastructure
Deliverable
production ML models
Required skills
ML compiler stack development, graph compilers (Inductor, XLA, TVM), kernel DSLs (Triton, Helion, cuTile), algebraic rewriting, fusion, scheduling languages, loop-level optimisation, autotuning, graph capture, symbolic and dynamic shapes, control flow, automatic differentiation, GPU kernel programming, CUDA, tensor cores, static analysis, SSA, CFGs, instruction selection, CUDA Graphs, KV caching, batching, parallelism strategies (Tensor, Sequence, Expert, Pipeline, FSDP), NCCL, activation checkpointing, CUDA-GDB, Nsight Systems, Nsight Compute, proof assistants, automated theorem provers, embedded DSLs, staging, metaprogramming, type systems, functional programming
Preferred skills
Experience with high-level ML frameworks (PyTorch, Jax, TensorFlow)
Technologies
PyTorch, Jax, TensorFlow, Inductor, XLA, TVM, Triton, Helion, cuTile, CUDA, CUDA-GDB, Nsight Systems, Nsight Compute
Responsibilities
Shape the development of the compiler stack, implement optimisations, develop frontends, program GPU kernels, debug and optimise GPU code, integrate formal methods, design programming languages and DSLs
Seniority
Senior, hands-on IC
