Anthropic25.07.2026
Research Scientist, Takeoff Intel
Зарплата не указана
РЫНОК
15 958 ₽медиана по профессии
Data Scientist · 148 вакансий с указанной зарплатой
5 000половина предложений: 12 500–22 000260 000
Работодатель не указал зарплату — сравните с рынком сами.
San Francisco
Обязанности
- 01Identify the signals that track AI R&D acceleration and design the evaluations that measure them
- 02Build quantitative models of capability growth and self-improvement dynamics, grounded in evaluation and telemetry data
- 03Run experiments and evals to test hypotheses about automation and capability
- 04Make opinionated research bets and own the outcome
- 05Write graded assessments of what our measurements show, for internal decision-makers and public reporting
- 06Collaborate with pretraining, RL, economic research, and policy teams
Требования
- 01Have done hands-on research on large language models: pretraining, fine-tuning, RL, evals, or agent systems
- 02Have strong quantitative instincts, are comfortable with quantitative modeling and reasoning
- 03Have experience in forecasting, may have published AI forecasting scenarios
- 04Can design an evaluation from a vague question and defend the methodology
- 05Write clearly and calibrate: state confidence, name what would change your conclusion
- 06Are motivated by impact: comfortable with work whose output is graded assessments and system-card sections more often than papers
- 07Care about AI safety and think carefully about where rapid capability growth leads
- 08Minimum education: Bachelor's degree or an equivalent combination of education, training, and/or experience
- 09Required field of study: A field relevant to the role as demonstrated through coursework, training, or professional experience
Условия
- 01Annual Salary: $350,000 — $850,000 USD
- 02Location-based hybrid policy: Currently, we expect all staff to be in one of our offices at least 25% of the time
- 03Visa sponsorship: We do sponsor visas
- 04We encourage you to apply even if you do not believe you meet every single qualification