AI Infra Performance Engineer — Experiments & Metrics

Commonai C.I.C.; Cambourne; Hybrid; GBP 70,000 - 90,000; Full time

Benefits offered by this job

Collaborative environment

High impact in growing org

Competitive salary and pension

Professional development

Networking with tech & academia

Office near Cambridge train station

Job summary

CommonAI CIC is seeking a skilled researcher to design and run experiments that reveal how AI infrastructure behaves, where bottlenecks lie and how observations can calibrate simulation models. You will build measurement tools, collect traces and counters, and work with workload, simulation and uncertainty specialists to resolve important unknowns.

Hybrid work with regular time in Cambridge; the team values deep expertise across hardware and software stacks and offers opportunities to broaden

Qualifications

• Experience investigating the performance of real computing systems.
• Depth in accelerator profiling, inference runtimes, memory systems, distributed/HPC workloads, networking, or power and energy measurement.
• Ability to build maintainable automation and analysis tools in a suitable software language, and to learn unfamiliar profiling tools.
• Knowledge and care of experimental controls, repeatability, confounding effects and the limits of a measurement.
• Some experience using AI coding agents, alongside debugging and review habits, is useful.

Responsibilities

• Design and run reproducible experiments from components to full LLM-serving systems.
• Develop reusable measurement tools and collect traces, profiles and counters to characterise latency, throughput and memory behaviour.
• Investigate bottlenecks, performance variation and unexpected behaviour across stack layers.
• Ensure reproducible workload details, software versions, configuration and measurement methods.
• Collaborate with simulation and uncertainty specialists to select measurements and curate data.
• Coordinate access and changes with colleagues responsible for hardware controls.

Skills

Performance analysis

Experiment design

Profiling tools

Automation tooling

LLM workloads

Tools

Nsight

PyTorch Profiler

System profilers

Job description

CommonAI CIC is seeking a skilled researcher to design and run experiments that reveal how AI infrastructure behaves, where bottlenecks lie and how observations can calibrate simulation models. You will build measurement tools, collect traces and counters, and work with workload, simulation and uncertainty specialists to resolve important unknowns.

Hybrid work with regular time in Cambridge; the team values deep expertise across hardware and software stacks and offers opportunities to broaden