Research Engineer, Agents

About Anthropic

Anthropic’s mission is to create reliable, interpretable, and steerable AI systems. We want AI to be safe and beneficial for our users and for society as a whole. Our team is a quickly growing group of committed researchers, engineers, policy experts, and business leaders working together to build beneficial AI systems.

About the role:

Agentic systems are becoming an increasingly important part of how AI is deployed.  Over the last year, we’ve seen rapid adoption of Claude-powered agentic systems in spaces like coding, research, customer support, network security, and more. We believe this is just the beginning, and we expect Claude to be handling much more complex tasks end-to-end or in cooperation with a human user as time goes on.  We have a team striving to make Claude an even more effective agent over longer time horizon tasks, and coordinate with groups of other agents at many different scales to accomplish large tasks. This team endeavors to maximize agent performance by solving challenges at whatever level is needed, whether it’s novel harness design, improved agent affordances and infrastructure, or finetuning.

Given that this is a nascent field, we ask that you share with us a project built on LLMs that showcases your skill at getting them to do complex tasks. Here are some example projects of interest: design of complex agents, quantitative experiments with prompting, constructing model benchmarks, synthetic data generation, or model finetuning. There is no preferred task; we just want to see what you can build. It’s fine if several people worked on it; simply share what part of it was your contribution. You can also include a short description of the process you used or any roadblocks you hit and how to deal with them, but this is not a requirement.

Responsibilities:

• Ideate, develop, and compare the performance of different agent harnesses  (eg memory, context compression, communication architectures for agents)
• 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 our product org to find solutions to our most vexing challenges applying agents to our products
• Help create and optimize data mixes for model training that maximize Claude’s performance or ease of use on agentic tasks

You may be a good fit if you:

• Have experience developing complex agentic systems using LLMs
• Have significant software engineering and ML experience
• Have spent time prompting and/or building products with language models
• Have good communication skills and an interest in working with other researchers on difficult tasks
• Have a passion for making powerful technology safe and societally beneficial
• Stay up-to-date and informed by taking an active interest in emerging research and industry trends.
• Enjoy pair programming (we love to pair!)

Strong candidates may also have experience with:

• Large-scale RL on language models
• 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 a novel eval that measures how many agents interact in groups to solve problems
• Build a scaled model evaluation framework driven by model-based evaluation techniques.
• Build the prompting and model orchestration for a production application backed by a language model
• Finetune Claude to maximize its performance using a particular set of agent tools or harness