Safeguards Enforcement Analyst, Safety Evaluations

Written by the T&S Network from Anthropic's posting.

Enforcement and safety evaluation work for an AI model developer's Safeguards team, checking that models meet safety and policy standards before and after launch. The role runs and monitors evals, drives mitigations when issues surface, and builds processes and tooling so the eval program can scale across the product surface as it expands.

What you'd do

• Run and monitor safety evaluations to support model launch readiness
• Surface regressions and unexpected model behavior changes to relevant stakeholders
• Scope evaluation approaches with policy and domain experts from risk identification onward
• Coordinate creation of new evals and keep existing ones current with evolving policies and threat vectors
• Build processes and frameworks for product-specific evaluations as the product surface grows
• Help design tooling that lets non-technical users create and iterate on evals self-serve
• Write and maintain documentation for eval creation, execution, and interpretation

What it asks for

• Operational trust and safety, content moderation, or policy enforcement background at a tech company
• Experience building processes or programs from scratch
• Strong program management skills across multi-stakeholder efforts
• Ability to manage multiple workstreams and make judgment calls with incomplete information
• Clear written and cross-functional communication
• Bachelor's degree, or equivalent skills gained through training or work experience
• Proficiency with SQL, dashboards, and spreadsheets (preferred)

Worth knowing

• Work involves comfort with sensitive content areas as part of eval creation or enforcement review

Our best summary of the posting, and it can be wrong. The employer's posting is what counts: read it in full on Anthropicnorth_east before you apply. If we got something wrong, report this listing.