Senior CUDA Engineer - High-Performance GPU Inference

Fuse Energy, LLC; Greater London; On-site; GBP 120,000 - 180,000; Full time

Benefits offered by this job

Equity sign-on bonus

Fully expensed tech equipment

Breakfast and dinner allowance

Biannual bonus

Job summary

Fuse Energy is seeking a CUDA Engineer to design and optimize low-level GPU kernels powering our transformer inference workloads. You will write custom CUDA kernels, tune memory bandwidth, and push the throughput of our GPU fleet at the SM, warp, and memory hierarchy level.

Collaborate with ML and systems teams to profile, fuse kernels, implement quantisation-aware arithmetic, and ensure robust tests and documentation.

Qualifications

• 4+ years writing production CUDA code for performance-critical kernels.
• Deep understanding of GPU microarchitecture, warps, occupancy and memory hierarchy.

Responsibilities

• Write and optimise custom CUDA kernels for core transformer inference operations.
• Profile kernels to identify and eliminate bottlenecks in occupancy, memory throughput, and warp divergence.
• Apply kernel fusion to reduce memory round-trips and launch overhead across inference pipelines.
• Optimise memory access patterns and manage the memory hierarchy for maximum bandwidth utilisation.
• Implement quantisation-aware kernels and mixed-precision arithmetic to reduce latency and memory footprint.
• Build and tune caching mechanisms for efficient autoregressive decoding.
• Tune kernel launch configurations for target GPU architectures.
• Benchmark kernels against existing baselines and drive measurable throughput and latency improvements.
• Write tests for CUDA code to catch performance and correctness regressions.
• Maintain internal CUDA libraries and contribute to team coding standards and documentation.

Skills

CUDA programming

GPU performance tuning

Profiling and debugging

Kernel fusion

Quantised/mixed-precision

Memory hierarchy optimization

CUDA C++

Job description

Fuse Energy is seeking a CUDA Engineer to design and optimize low-level GPU kernels powering our transformer inference workloads. You will write custom CUDA kernels, tune memory bandwidth, and push the throughput of our GPU fleet at the SM, warp, and memory hierarchy level.

Collaborate with ML and systems teams to profile, fuse kernels, implement quantisation-aware arithmetic, and ensure robust tests and documentation.