Research Engineer – Human Influence

AI Security Institute; Greater London; On-site; GBP 70,000 - 110,000; Full time

Job summary

AI Security Institute is seeking a Research Engineer to join the Human Influence team. You will contribute to scalable LLM post-training and reinforcement learning projects, and help evaluate model performance and safety across large compute resources.

The role emphasizes building robust evaluation pipelines, applying interpretability methods, and delivering engineering-heavy research on AI risk and human influence topics. A multidisciplinary background is valued.

Qualifications

• Clear understanding of the current AI safety literature and Human Influence topics.
• Proven ability to deploy a benchmark/evaluation pipeline to users or scale research artifacts.
• Excellent written and verbal communication skills.

Responsibilities

• Designing and implementing scalable RL and post-training workflows for LLMs.
• Using interpretability tools to analyze model behavior and mitigation strategies.
• Building and maintaining evaluation pipelines and benchmarks across model releases.

Skills

Communication

ML fundamentals

Benchmark/evaluation deployment

Tools

PyTorch

Keras

JAX

Job description

About the AI Security Institute

The AI Security Institute is the world's largest and best-funded team dedicated to understanding advanced AI risks and translating that knowledge into action. We’re in the heart of the UK government with direct lines to No. 10 (the Prime Minister's office), and we work with frontier developers and governments globally.

We’re here because governments are critical for advanced AI going well, and UK AISI is uniquely positioned to mobilise them. With our resources, unique agility and international influence, this is the best place to shape both AI development and government action.

The deadline for applying to this role is Sunday 11th October 2026, end of day, anywhere on Earth.

TeamDescription

The Human Influence (HI) team focuses on the ways in which AI can influence human beliefs, decisions, and behaviour. A substantial class of AI risk operates through people. AI systems can persuade people to change their beliefs and to take action; can build trusting relationships with people in order to exploit them; can extract private information from them; and can hold delegated ownership of high-stakes decisions.

Our work is highly interdisciplinary, drawing on methods from computational social science, AI safety and security, cognitive and behavioural science, machine learning, and data science. Typical projects include running rigorous human-AI interaction studies and randomised controlled trials, building evaluations and benchmarks that track AI capabilities across model releases, eliciting model capabilities through fine-tuning and self-play, and developing datasets to monitor real-world risk exposure and severity.

Role Description

We are looking for a Research Engineer to join the Human Influence team. Successful candidates will be strong researchers and engineers with a track record of carrying out scalable work in LLM post-training and fine-tuning, especially with Reinforcement Learning; or with comparable expertise in engineering and validating large-scale evaluation pipelines.

Projects the Research Engineer might deliver include:

• Designing and building a Reinforcement Learning environment aimed at mitigating a model’s ability to e.g. deceive a user in a one-to-one conversation or within multi-agent threads.
• Leveraging state-of-the-art interpretability methods to identify why models exhibit concerning behaviour, and designing mitigations that can be applied to models irrespective of training regime.
• Building the scalable system architecture underpinning the repeatable delivery and analysis of model evaluations and benchmarks.
• Delivering ambitious, engineering-heavy research projects on Human Influence topics, for instance by leveraging post-training techniques on a large compute cluster.

Who we're looking for

This is a multidisciplinary team, and successful candidates come from a wide range of backgrounds.

Essential

General:

• Proven experience deploying a benchmark, evaluation, or product to users, e.g. an evaluations pipeline in an app, an open-source contribution, or similar large-scale contributions to research.
• Clear understanding of the current AI safety literature, and an interest in topics relevant to Human Influence.
• Clear and consistent communication.
• Clear understanding of fundamental Machine Learning concepts.

Research and engineering:

• Experience fine-tuning or post-training LLMs using standard methods, using common libraries like PyTorch, Keras, JAX, or custom code.