Research Engineer/Research Scientist - Red Team (Misuse)

GBP 29K-145K Mid-level Full Time Found 11d ago

Tasks

• Build scalable, reliable research code
• Collaborate with frontier AI companies and government agencies
• Contribute to AI safety and security projects
• Create benchmarks for monitoring misuse and jailbreaks
• Design, build, run, and evaluate automated attack systems
• Develop and analyse attacks and protections for large language models
• Investigate attacks and defenses for data poisoning and backdoors
• Perform adversarial testing of AI safeguards
• Publish impactful research in top-tier venues

Perks/Benefits

• Autonomy in research
• Conference funding
• Generous leave
• Hybrid work options
• Impactful work
• Lifestyle discounts
• Operational support
• Parental leave
• Pension contributions
• Pre-release models and compute access
• Work with top AI firms and government

Skills/Tech-stack

AI Security | Adversarial Robustness | Automation | Code building | Data poisoning | Evaluation Frameworks | Experiment design | Language Models | Large Language Models | ML frameworks | PyTorch | Reinforcement Learning | Reproducibility | Research coding