Research Scientist/Engineer – Biological Safety

About the role

Anthropic is seeking a Research Scientist/Engineer to develop safety and oversight mechanisms for biological knowledge in its AI models. You will design evaluations, create training data, and build classifiers that protect against misuse while supporting legitimate scientific work.

Key responsibilities

• Design, build, and run capability evaluations for biological domains and translate results into deployment recommendations.
• Develop training and evaluation datasets for safety classifiers in collaboration with threat‑modeling experts.
• Train, tune, and iterate on classifiers with ML engineers, focusing on adversarial robustness and low false‑positive rates.
• Create tooling and pipelines to make evaluation and classifier development fast and repeatable.
• Analyze classifier performance against production traffic, identify gaps, and prioritize improvements.
• Conduct red‑team and stress‑testing of safeguards as models and product surfaces evolve.
• Partner with Research, Product, and Policy teams to embed biological safety throughout the model lifecycle.
• Contribute to external communications such as model cards, blog posts, and policy documents.

Required profile

• Strong scientific background in modern biology.
• Proficiency in Python and extensive experience with scientific programming and data analysis.
• Solid understanding of machine‑learning fundamentals and ability to adopt new ML techniques quickly.

Required skills

• Python