Staff ML Performance Engineer (Compiler)
On-siteLondon, England, United Kingdom
Job Summary
Optimize ML inference for edge accelerators and GPUs by implementing and validating compiler, runtime, and kernel improvements such as operator fusion, scheduling, and quantization-aware performance. Profile the full inference stack to pinpoint bottlenecks and deliver measurable latency and memory reductions across NVIDIA Orin/Thor and Qualcomm platforms. Build robust benchmarking and regression testing to ensure performance stability across models, devices, and software releases. Collaborate with model developers to influence architecture and deployment decisions while contributing to technical roadmaps and team-wide performance standards.
Required Qualifications
- Proven experience improving performance in production systems with tight constraints (latency, memory, bandwidth, power/thermal, or cost)
- Strong proficiency with at least one relevant stack/toolchain (e.g. TensorRT, CUDA, Qualcomm QNN, Triton, OpenCL, MLIR, ONNX) and confidence learning adjacent frameworks quickly
- Comfort operating at multiple levels of abstraction — from high-level model behaviour down to low-level kernel/runtime execution
- Strong software engineering fundamentals (debugging, profiling, testing, and maintainable code)
- Clear communicator and collaborative teammate; able to align multiple stakeholders on performance trade-offs and priorities
Desired Qualifications
- Experience with compute graph scheduling and execution on multiple targets
- Exposure to embedded or edge deployment of ML models, including benchmarking on real devices and handling system-level constraints
- Experience with NVIDIA and/or Qualcomm SoCs and performance tooling
- Python and C++ proficiency
- Experience mentoring others and/or driving technical direction in a small, fast-moving team
Hiring someone like this?
Get your role in front of qualified candidates on Sorce.