Anthropic
Research Scientist/Engineer, Biological Safety
About this role
Anthropic seeks a research scientist/engineer to develop safety mechanisms and oversight systems for how AI models handle biological knowledge. You'll design capability evaluations, build training datasets for safety classifiers, and measure their effectiveness against adversarial threats while balancing robust safeguards with legitimate research access.
What you'll do
- Design and execute capability evaluations to assess biological domain performance of frontier models and translate findings into deployment recommendations
- Develop and curate training and evaluation datasets for safety classifiers grounded in realistic threat models
- Train, tune, and optimize safety classifiers for adversarial robustness and low false-positive rates alongside ML engineers
- Build tooling and pipelines to enable fast, repeatable evaluation and classifier development workflows
- Analyze production classifier performance, identify safety gaps, and prioritize iterative improvements
- Conduct red-teaming and stress-testing of biological safeguards as threats and models evolve
What they're looking for
- Advanced Python and scientific programming
- Machine learning fundamentals and development practices
- Modern biology (assays, genome editing, protein engineering, gene synthesis)
- Experimental design and quantitative analysis
- Biosecurity frameworks and dual-use research concerns
- Technical communication and cross-functional collaboration
- LLM evaluation and adversarial robustness
Opens the application — the Jobs AI extension fills it for you. Set up autofill
Opens the official application on the employer’s site. No login required.
Anthropic
Anthropic builds Claude, an AI assistant, and is hiring for engineering roles across infrastructure, data systems, and security that support both AI research operations and the company's internal technology needs. The company seeks infrastructure engineers, systems integrators, data scientists, and security specialists to build production-scale systems for training data pipelines, financial operations, developer productivity measurement, research infrastructure, and server firmware security.
- Website
- anthropic.com
Likely interview questions
- Walk us through how you would design an evaluation to assess whether a model can generate dangerous biological instructions, and how you'd balance false positives versus true positives.
- Describe your experience developing datasets for ML systems and how you've ensured they capture realistic edge cases or adversarial scenarios.