AI Evaluation Engineer

Incubator for Artificial Intelligence; Manchester; Hybrid; GBP 46,000 - 60,000; Full time

Benefits offered by this job

Cycle to work

Dental insurance

Hybrid work arrangement

25+ days leave

Job summary

The Incubator for AI (i.AI) is hiring two AI Evaluation Engineers to focus on model-level evaluation of accuracy, bias and safety. You will design evaluation pipelines and work with product and engineering teams to embed evaluation from day one through to scale, ensuring responsible AI across public services.

You'll turn findings into actionable recommendations, collaborate with domain experts on high-quality datasets, and help stakeholders understand real-world impact.

Qualifications

• Experience in a scientific, analytical or engineering role.
• Ability to design experiments and interpret results.
• Experience evaluating AI/ML systems.
• Strong communication of technical findings to diverse audiences.
• Proficiency in Python and version control (Git).

Responsibilities

• Designing and building evaluation pipelines and experiments for AI products.
• Running model and system evaluations to test accuracy, bias, safety and overall performance.
• Working with product and engineering teams to integrate evaluation from day one through scale.
• Partnering with domain experts to create and maintain high-quality golden datasets and benchmarks.
• Turning evaluation results into clear, practical recommendations for product teams.
• Helping teams understand the real-world impact of their products on public services.
• Communicating findings to both technical and non-technical audiences through reports and presentations.
• Developing reusable evaluation approaches that work across many products and services.

Skills

Analytical thinking

Python

Git

Experimental design

AI/ML evaluation

Communication

Tools

Git

Job description

AI Evaluation Engineer (Data Scientist) - London/Manchester/Bristol

The Incubator for AI (i.AI) is a team of around 80 builders and operators using frontier AI to improve public services across healthcare, education, planning and parliament. We work at startup pace, with the backing of the Prime Minister and No10, on problems that affect millions of people.

Our Impact and Evaluation team embeds within product teams to build rigorous evaluation into every stage of development, shaping product decisions and showing whether our products improve public services. We're now hiring two AI Evaluation Engineers to focus on model-level evaluation of accuracy, bias and safety.

What will I be doing?

• Designing and building evaluation pipelines and experiments for AI products
• Running model and system evaluations to test accuracy, bias, safety and overall performance
• Working with product and engineering teams to build evaluation in from day one through to scale
• Partnering with domain experts to create and maintain high-quality golden datasets and benchmarks
• Turning evaluation results into clear, practical recommendations for product teams
• Helping teams understand the real-world impact of their products on public service delivery
• Communicating findings to technical and non-technical audiences through reports and presentations
• Developing reusable evaluation approaches that work across many products and services

What will I need?

You don't need to tick every box. You'll likely bring some of the following:

• Experience in a scientific, analytical or engineering role
• Strong analytical and critical thinking skills, and the ability to draw insight from complex data
• Practical knowledge of experimental design and statistical hypothesis testing
• Proficiency in Python and version control (e.g. git)
• Experience evaluating generative or predictive AI/ML systems
• Experience working in a software development team
• The ability to explain technical work clearly to a broad audience

What's in it for me?

• Projects backed by No10 that apply AI to public services affecting millions, alongside talented, mission-driven colleagues
• Frontier models, ample compute, access to experts in national security, policy and AI research, and full operational and delivery support
• Early ownership of important products in small teams, 5 L&D days a year, and funding for learning, conferences and collaborations
• £48,540-£62,970 in London (£45,740-£59,835 nationally), including a technical allowance of up to £9,000, plus a 28.97% employer pension contribution
• Flexibility and benefits. Hybrid working from London, Manchester or Bristol, 25+ days' leave, up to 39 weeks' full-pay parental leave, and perks including cycle to work and dental insurance