About the role
Every time Anthropic releases a model, we publish a system card: a long-form technical document that describes the model's capabilities, safety properties, evaluation results, and the reasoning behind our deployment decisions. System cards are some of the most consequential and widely read documents we produce, and they are one way we hold ourselves publicly accountable for the safety claims we make.
We're hiring a Research Operations Specialist to help own system card production. You'll work embedded with research and safety teams through each launch, coordinating contributions from dozens of researchers, holding the schedule and the open-threads list, and making sure the document ships on time as a single, accurate, internally consistent whole. Along the way you'll do real editorial work: turning results and researcher notes into clear, honest prose and pushing back when an explanation doesn't hold together.
System cards sit within a wider family of external safety artifacts, including risk reports and Responsible Scaling Policy updates. Part of this role is keeping the system card consistent with those documents so that Anthropic's public safety story reads as one coherent account rather than several.
This role sits in Research Operations and works closely with Alignment, Safeguards, Frontier Red Team, and capabilities research. The core of the job is part project management, part translation: keeping a complex, many-author, hard-deadline document on track while making frontier safety research legible to researchers, policymakers, journalists, and the public — without sacrificing precision.
Key responsibilities:
- Drive system card production end to end — own the timeline, the contributor list, the open-threads tracker, and the definition of done for each launch
- Coordinate dozens of contributors across Alignment, Safeguards, Frontier Red Team, Interpretability, and capabilities; chase drafts, resolve differences of perspective, seek ground truth, and run final document beautification
- Edit, and sometimes write, content; work directly with researchers to turn their results, notes, and plots into clear, scientific, non-marketing prose and maintain Anthropic's voice across sections drafted by many different people
- Guard accuracy and consistency; catch terminology drift, claims that subtly contradict each other, and discrepancies between internal communications and system card drafts
- Keep the system card aligned with related artifacts; track what's being said in risk reports, RSP disclosures, and other safety documentation so the system card stays consistent with them, and flag conflicts early
- Improve the process between launches; build and maintain templates, style guidance, contributor guides, checklists, and reusable section scaffolds so each cycle starts from a stronger baseline
- Pick up other research-adjacent operations and writing, such as internal research summaries, release notes, and documentation that helps research leadership communicate clearly
Minimum qualifications:
- Strong project management and execution skills; able to track dozens of open threads to closure during compressed launch periods with many moving parts and many owners
- Demonstrated ability to coordinate and influence without direct authority, building trust while chasing drafts and giving constructive editorial feedback
- Excellent technical writing skills; able to take dense, jargon-heavy source material and produce prose that is precise, honest, and readable by a smart non-specialist
- Comfort working closely with researchers; able to read an evaluation results table, ask the right clarifying questions, and push back when an explanation doesn't hold together
- Working knowledge of large language models at a conceptual level, including fluency with vocabulary such as pretraining, RLHF, context windows, evals, and red-teaming
- Data communication literacy; able to identify how a chart or table could be made clearer and more accurate
- High integrity and a genuine sense of accountability around producing documents that hold Anthropic publicly responsible for its safety claims
Preferred qualifications:
- Familiarity with AI safety, AI policy, alignment research, evaluation methodology, or the RSP landscape beyond baseline LLM knowledge
- Background in science communication, research publishing, or technical journalism
- A track record of shipping long-form technical documents: research reports, whitepapers, technical standards, regulatory filings, or science journalism
- Experience with safety, risk, or compliance documentation such as regulatory submissions, safety cases, security disclosures, or clinical/scientific reporting
- An eye for visual communication and experience shipping polished, on-brand documents
- Light scripting or tooling ability (enough to automate a formatting check or wrangle a spreadsheet)
