Performance Engineer: HPC & AI Optimization, Equity

CuspAI; Cambridge; Hybrid; GBP 90,000 - 130,000; Full time

Benefits offered by this job

Competitive salary

Equity in CuspAI

Holiday: 28 days (DE, NL, UK)

Parental leave: 26 weeks

Professional development budget

Interdisciplinary teamwork

Job summary

CuspAI is seeking a Performance Engineer in the UK to bridge frontier AI models with high‑performance computing infrastructure. You will profile training runs, identify bottlenecks in MFU, and design custom kernels for larger, more complex models in chemistry and physics.

You will collaborate with AI researchers and the Infrastructure team to shape future compute clusters, optimize data pipelines, and implement efficiency techniques to reduce latency.

Qualifications

• 3+ years of experience in performance engineering, site reliability engineering, or low‑level software development, focused on HPC or AI workloads.
• Expert knowledge of GPU architecture and CUDA kernel optimization.
• Deep understanding of distributed training frameworks (e.g. PyTorch Distributed, DeepSpeed, Megatron‑LM) and networking protocols.
• Proficiency with profiling tools such as Nsight Systems, Nsight Compute, or Weights & Biases for performance tracking.
• Autonomous, builder mindset suitable for an early‑stage startup environment.

Responsibilities

• Analyze model architectures to optimize operations mapping onto hardware, focusing on NVIDIA GPU kernels and NCCL.
• Implement efficiency techniques like quantization, pruning, and distillation to lower latency for AI workloads.
• Profile and optimize data pipelines to prevent data starvation during large training runs.
• Use observability tools to trace end‑to‑end training execution and pinpoint bottlenecks.
• Build dashboards tracking MFU, memory bandwidth, and cluster uptime; justify AI accelerator utilization.

Skills

Performance engineering

Site reliability

Low‑level software

GPU architecture

CUDA kernels

Triton

PyTorch Distributed

DeepSpeed

Megatron‑LM

Profiling tools

Nsight Systems

Nsight Compute

Weights & Biases

Distributed training

Networking protocols

Builder mindset

Tools

NVIDIA GPUs

NCCL

CUDA

Job description

CuspAI is seeking a Performance Engineer in the UK to bridge frontier AI models with high‑performance computing infrastructure. You will profile training runs, identify bottlenecks in MFU, and design custom kernels for larger, more complex models in chemistry and physics.

You will collaborate with AI researchers and the Infrastructure team to shape future compute clusters, optimize data pipelines, and implement efficiency techniques to reduce latency.