ML Systems Engineer: Inference Benchmarks, Hybrid Cambridge

CommonAI Holdings Ltd; Cambridge; Hybrid; GBP 70,000 - 110,000; Full time

Benefits offered by this job

Pension

Professional development opportunities

Hybrid work model

Office near Cambridge train station

Job summary

CommonAI CIC seeks a dedicated engineer to build reproducible LLM serving workloads and benchmarks for long-context and agentic tasks. You will describe workload behavior, define performance metrics, and document configurations and provenance to enable reproducible comparisons across systems.

The role blends measurement and simulation work in a collaborative, award-winning research setting. Hybrid role with Cambridge attendance, open to UK applicants, offering a substantial tooling budget and

Qualifications

• Experience building or evaluating ML systems, inference services, benchmarks or workload-driven systems.
• Strong software skills for reliable benchmark tools and result analysis.
• Understanding of LLM inference and how serving choices affect performance and output quality.
• Experience with AI coding agents and validating generated code and data transformations.
• Depth in vLLM or SGLang or similar serving software is valuable.

Responsibilities

• Build reproducible LLM serving workloads and benchmark tools for long-context and agentic workloads.
• Characterise request arrivals, prompt/output lengths, caching, batching and tool interactions.
• Define performance measures aligned with application quality and service requirements.
• Coordinate with measurement and simulator engineers to reproduce workloads on physical systems.
• Keep benchmarks relevant as inference tech and workloads evolve.
• Document datasets, model/software versions, configurations, provenance and usage terms.
• Contribute to comparisons across systems under agreed requirements.

Skills

ML systems

Benchmark tools

LLM inference

Software debugging

vLLM

SGLang

Trace analysis

Education

Experience-focused, non-degreed acceptable

Tools

vLLM

SGLang

Job description

CommonAI CIC seeks a dedicated engineer to build reproducible LLM serving workloads and benchmarks for long-context and agentic tasks. You will describe workload behavior, define performance metrics, and document configurations and provenance to enable reproducible comparisons across systems.

The role blends measurement and simulation work in a collaborative, award-winning research setting. Hybrid role with Cambridge attendance, open to UK applicants, offering a substantial tooling budget and