Research Manager, Biological Safety

Anthropic•San Francisco, CA

8d•Hybrid

About The Position

Anthropic's Safeguards organization is responsible for developing policies, evaluations, and enforcement systems to prevent AI models from causing catastrophic harm. This role involves managing a research engineering team focused on biological safety. The team's work includes designing and executing capability evaluations for advanced AI models, curating training data for safety classifiers, training and refining these classifiers in collaboration with ML engineers, and assessing their performance against adversarial pressures in live traffic. The manager will define the technical strategy, prioritize team efforts, and be accountable for outcomes. This is a hands-on management position requiring significant time dedicated to team growth and direction, while maintaining enough technical expertise to review evaluation designs, analyze classifier failure modes, and effectively communicate with Research, Product, and Policy teams. The team's primary challenge is balancing the need for robust safeguards against sophisticated actors with the goal of not impeding legitimate research by the broader community using AI for life sciences. This tradeoff is an empirical problem that the team will measure and address.

Requirements

Experience managing a technical team, including hiring, coaching, and performance management.
A record of setting technical direction for a team and making prioritization calls under uncertainty.
Proficiency in Python, with a background in scientific programming and data analysis.
A solid grasp of ML fundamentals, sufficient to critically review evaluation design and classifier development.
Knowledge of modern biology across both measurement and engineering: high-throughput assays and functional characterization, as well as gene synthesis, genome editing, strain construction, and protein engineering.
Experience designing quantitative experiments or evaluations and drawing defensible conclusions from noisy results.
Clear analytical and writing skills, and the ability to explain technical concepts to non-technical stakeholders.
Familiarity with dual-use research concerns and biosecurity frameworks, such as select agent regulations, the Biological Weapons Convention, or Australia Group guidelines.
Comfort with ambiguity and with shifting priorities as AI capabilities change.
Motivation to prevent misuse without obstructing the beneficial work that makes up the vast majority of this field.

Nice To Haves

3+ years of people management experience, ideally leading research scientists, research engineers, or ML engineers.
Experience building a team or function from a small headcount, including defining scope, hiring the first few people, and establishing how the team works.
At least 8 years of hands-on experience in life sciences, with deep expertise in areas such as molecular biology, drug discovery, or computational biology.
Experience working with large language models, including prompting, fine-tuning, or evaluation.
Experience training or deploying classifiers or other ML systems in production, and comfort reasoning about precision and recall for rare, high-consequence categories where the base rate is very low.
Experience developing ML methods for biological systems or biological data.
Familiarity with adversarial robustness, red-teaming, or safety evaluation of ML systems.
Experience leading complex technical projects across multiple stakeholder groups.

Responsibilities

Manage, coach, and grow a team of research scientists and engineers working on biological safety evaluations and classifiers, including hiring, onboarding, performance, and career development.
Set the technical direction and roadmap for the biological safety research agenda, and make the calls about what the team builds, what it deprioritizes, and when a safeguard is ready to ship.
Own the quality of capability evaluations that assess what new models can do in the biological domain, and turn results into deployment recommendations that leadership can act on.
Guide the development of training and evaluation datasets for our safety classifiers, working with internal and external threat modeling experts to ground them in realistic risk.
Oversee the training and iteration of safety classifiers alongside ML engineers, optimizing jointly for adversarial robustness and low false-positive rates.
Ensure the team invests in the tooling and pipelines that make evaluation and classifier development fast and repeatable.
Establish how the team measures classifier and eval performance against production traffic, identifies gaps, and prioritizes improvements.
Direct red-teaming and stress-testing of safeguards as threats, models, and product surfaces evolve.
Partner with Research, Product, Policy, and government affairs colleagues to embed biological safety throughout the model development lifecycle, and serve as an escalation point for biological content.
Represent the team's work in external communications including model cards, blog posts, and policy documents.
Track developments in biology, machine learning, and biosecurity for their potential to create new risks or enable new mitigations.