Anthropic Research Engineer, Domain Scaling: AI Research Job in the United States Paying Up to $850,000
Overview
Anthropic is hiring a Research Engineer, Domain Scaling, to help improve Claude on complex professional knowledge work. The team concentrates on fields such as finance, healthcare and legal services. The post is full-time and based in the United States, with offices in San Francisco, New York City and Seattle. The advertised annual salary runs from USD $350,000 to $850,000, depending on experience, qualifications and internal job level.
The job blends applied AI research, engineering, data strategy and reinforcement learning. The engineer would own data strategies for knowledge-work areas, from identifying valuable tasks and sourcing real-world and synthetic data through to reinforcement learning training. Other duties include managing technical relationships with external data vendors, judging data quality, contributing to reward design, working with domain experts on data pipelines and evaluation systems, and building quality-assurance frameworks that catch reward hacking. The engineer would also run generalisation experiments to see how data strategies affect model capabilities and turn capability goals into training environments and evaluations.
Who can apply
The company wants people with strong experience in AI and machine learning, such as large language models, reinforcement learning or LLM training data. Fine-tuning models for specific domains, machine learning research and experimentation, data curation, reward design, data quality assurance and technical data sourcing all count in an application. Experience training production machine learning systems, building LLM evaluations or benchmarks, or working with external research and technology partners may make a candidate more competitive.
The minimum education requirement is a Bachelor's degree or an equivalent mix of education, training and experience. Candidates need not meet every listed qualification.
What you get
Beyond salary, the benefits may include equity, donation matching, vacation and parental leave, flexible working arrangements and access to collaborative offices. The working model is hybrid, with at least 25% of time expected in an office, and some roles may need more. Visa sponsorship is available for eligible roles and candidates.
How to apply
Applications go through the company's official careers platform. Applicants should be ready to provide a resume or LinkedIn profile and describe why they want to join. The process may also ask about relocation, visa sponsorship, willingness to work from an office and possible start dates. No deadline is stated.