Agentic AI Research Engineer — Systems & Evaluation

Callosum; Greater London; On-site; GBP 120,000 - 190,000; Full time

Benefits offered by this job

Visa sponsorship

Relocation benefits

Private healthcare

London office access

Job summary

Callosum, based in London, is advancing the science of agentic AI through application-aware orchestration across heterogeneous models and hardware. The team builds rigorous tools to analyze why agentic systems fail and uses findings to guide design decisions.

You will design experiments, run analyses, and contribute to scalable infrastructure that supports the company’s evolving silicon and model stack. The role offers latitude to choose problems, with the opportunity to publish insights and

Qualifications

• Ability to design and run rigorous experiments for AI systems.
• Experience with long-horizon, multi-step tasks and tool use.
• Proven discipline in forming hypotheses and evaluating results.
• Strong software engineering to run experiments in a shared codebase.
• Clear, prioritised communication of research findings.

Responsibilities

• Design and run experiments to isolate failure modes in agentic systems.
• Contribute to core research themes: memory, task decomposition, coordination.
• Build analysis and evaluation infrastructure for reliability and cost.
• Publish evaluation results and methods as company assets.
• Collaborate across teams to scale experiments and reduce latency.

Skills

Research design

Experimentation

LLMs in agentic settings

Systems engineering

Technical communication

Job description

Callosum, based in London, is advancing the science of agentic AI through application-aware orchestration across heterogeneous models and hardware. The team builds rigorous tools to analyze why agentic systems fail and uses findings to guide design decisions.

You will design experiments, run analyses, and contribute to scalable infrastructure that supports the company’s evolving silicon and model stack. The role offers latitude to choose problems, with the opportunity to publish insights and