Member of Technical Staff, AI Researcher

About the role

Teams in healthcare and government need AI they can trust in real workflows, so this role works on the evaluations and feedback loops that make model behaviour measurable.

At a glance

What you will do

• Build benchmarks and evaluation methods for large language models, agents, multimodal models and reinforcement learning settings
• Create reward models, graders and programmatic verifiers that turn model behaviour into something you can measure
• Run controlled fine-tuning and reinforcement learning experiments to see how datasets change capability, generalisation and alignment
• Design ways to collect high-quality human data, from expert task design to annotation methods and quality signals

What they look for

• Comfortable both chasing an open research question and building the system to test it
• Care about how data structure, selection, attribution and quality shape what a model does
• Can design focused experiments and read ambiguous results into technical decisions
• Move quickly while keeping engineering quality and maintainability