Research Engineer, Agents

Details

Anthropic is seeking a Research Engineer to help develop agentic systems that enable Claude to handle increasingly complex tasks independently or in cooperation with human users and other agents. The role involves improving agent performance through novel harness designs, agent affordances, infrastructure, evaluation, and fine-tuning.

Applicants are asked to share a project built on large language models that demonstrates their ability to make models perform complex tasks. Relevant projects may include complex agents, quantitative prompting experiments, model benchmarks, synthetic data generation, or model fine-tuning.

Responsibilities

• Ideate, develop, and compare agent harnesses, including memory, context compression, and communication architectures.
• Design and implement rigorous quantitative benchmarks for large-scale agentic tasks.
• Assist with automated evaluation of Claude models and prompts across the training and product lifecycle.
• Work with the product organization to solve challenging problems involving the application of agents to Anthropic products.
• Help create and optimize data mixes for model training to maximize Claude's performance and ease of use on agentic tasks.

Requirements

• Experience developing complex agentic systems using large language models.
• Significant software engineering and machine learning experience.
• Experience prompting or building products with language models.
• Good communication skills and an interest in collaborating with other researchers on difficult tasks.
• Passion for making powerful technology safe and beneficial to society.
• Active interest in emerging research and industry trends.
• Willingness to participate in pair programming.
• Bachelor's degree or an equivalent combination of education, training, and experience.
• Education, training, or professional experience in a field relevant to the role.

Additional Qualifications

• Experience with large-scale reinforcement learning on language models.
• Experience with multi-agent systems.

Representative Projects

• Design and build a novel agent harness that outperforms existing agents on coding or knowledge-work benchmarks.
• Design and build agent affordances that unlock new capabilities for internal use and deployed products.
• Design and build an evaluation measuring how groups of agents interact to solve problems.
• Build a scaled model-evaluation framework driven by model-based evaluation techniques.
• Build prompting and model orchestration for a production application backed by a language model.
• Fine-tune Claude to maximize performance using a particular set of agent tools or harness.

Compensation

The annual salary range is $500,000–$850,000 USD.

Work Arrangement

This is a remote-friendly role with travel required. Staff are currently expected to work from one of Anthropic's offices at least 25% of the time, although some roles may require more office time.

Benefits

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

Anthropic sponsors visas, although sponsorship is not guaranteed for every role or candidate. The company makes reasonable efforts to obtain visas and works with an immigration lawyer.