🛠 Tech Stack

💼 About This Role

You'll join our new AGI safety product team to transform complex AI research into practical tools that reduce AI risks. You'll work closely with the CEO and product engineers to build monitoring for AI coding agents. This role offers significant ownership and the chance to directly impact real-world AI safety through empirical research and rapid iteration.

🎯 What You'll Do

• Systematically collect and catalog coding agent failure modes from diverse sources.
• Design and run experiments to test monitor effectiveness across failure modes.
• Build and maintain evaluation frameworks for monitoring capabilities.
• Iterate on monitoring approaches based on empirical results.

📋 Requirements

• 2+ years conducting empirical research with LLMs or AI systems.
• Experience with AI coding agents (using, designing, or developing).
• Strong Python programming skills.
• Experience designing and running experiments and analyzing results.

✨ Nice to Have

• Experience with AI evaluation frameworks like Inspect.
• Familiarity with AI safety concepts and agent-related risks.
• Experience fine-tuning language models or working with open-source models.

🎁 Benefits & Perks

• 💰 Market competitive salary (100k-180k GBP).
• 🏖️ Unlimited vacation and sick leave.
• 👶 Up to 6 months paid parental leave.
• 🍱 Daily meals and snacks provided.
• 📚 $1,000 yearly professional development budget.