Research Engineer, Takeoff Intel

Details

Anthropic’s Takeoff Intel team measures how much AI development is becoming AI-assisted. The team designs evaluations of AI research and development capabilities, builds internal telemetry, and develops quantitative methods to assess capability growth and support AI safety and situational awareness.

This generalist Research Engineer role focuses on evaluation infrastructure, large-scale data processing, and analysis tooling. The team is hiring at junior and senior levels.

Responsibilities

• Design, build, and run capability evaluations and measurement instruments at scale.
• Build data and analysis pipelines that transform large volumes of model outputs and telemetry into reliable metrics.
• Prototype new instruments quickly, validate them, and determine what to retain.
• Review and supervise AI-written code as part of the normal workflow.
• Collaborate with research scientists and partner teams to define what should be measured.
• Contribute to internal write-ups and public reporting.

Requirements

• Experience shipping an evaluation, data product, or research library end to end.
• Ability to prototype quickly and discard code when appropriate.
• Comfort handling messy, large-volume data without over-engineering.
• Experience running experiments on large language models, rather than only moving their outputs.
• Ability to work from vague research questions rather than detailed specifications.
• Strong communication skills and ability to collaborate closely with researchers.

Strong candidates may also have:

• Experience building evaluation harnesses or benchmark infrastructure for large language models.
• Experience with large-scale machine learning or data infrastructure, including self-driving, observability, or similar systems, alongside machine learning exposure.
• Experience building tools or libraries used by other researchers.
• A track record of identifying errors in AI-written code.

Examples of Work

• Anthropic ECI, an adaptation of the Epoch Capabilities Index used in recent system cards to measure capability acceleration.
• AI research and development capability assessments in Claude system cards.
• Data and research supporting When AI Builds Itself.

Education and Experience

• Minimum education: Bachelor’s degree or an equivalent combination of education, training, and experience.
• Required field of study: A field relevant to the role, demonstrated through coursework, training, or professional experience.
• Required years of experience vary according to the internal job level.

Benefits

Anthropic offers competitive compensation and benefits, optional equity donation matching, generous vacation and parental leave, flexible working hours, and office spaces for collaboration.