Software Development · San Francisco, CA

On-site

Software Engineer, Research Tools

Anthropic · San Francisco, CA · Posted Nov 8, 2025

This posting comes from Anthropic. Confirm the details on their site before applying.

Overview

Anthropic's research teams are pushing the boundaries of AI safety and capability research, and they need exceptional tools to do their best work. As a Software Engineer on the Research Tools team, you'll build the infrastructure and applications that enable our researchers to iterate quickly, run complex experiments, and extract insights from frontier AI systems.

Responsibilities

  • Build and maintain full-stack applications and infrastructure that researchers use daily to conduct experiments, collect feedback, and analyze results
  • Partner closely with research teams to understand their workflows, pain points, and requirements, translating these into technical solutions
  • Design intuitive interfaces and abstractions that make complex research tasks accessible and efficient
  • Create reusable platforms and tools that accelerate the development of new research applications
  • Rapidly prototype and iterate on solutions, gathering feedback from users and refining based on real-world usage
  • Take ownership of complete product areas, from understanding user needs through design, implementation, and ongoing iteration
  • Contribute to technical strategy and architectural decisions for research tooling
  • Mentor other engineers and help establish best practices for research application development
  • Building interfaces for collecting and managing human feedback on model outputs at scale
  • Creating experiment orchestration platforms that make it easy to launch, monitor, and analyze complex research runs
  • Developing visualization tools that help researchers understand model behavior and identify failure modes
  • Designing reusable components and frameworks that enable rapid development of new research applications
  • Building sandboxed execution environments for safely running AI-generated code

Requirements

  • Have 5+ years of software engineering experience with a strong focus on full-stack development
  • Excel at rapid iteration and shipping—you can move from concept to working prototype quickly
  • Have experience building tools, platforms, or infrastructure for technical users (engineers, researchers, data scientists, analysts, etc.)
  • Demonstrate high agency and ability to operate independently in ambiguous environments
  • Can quickly develop deep understanding of complex technical domains
  • Have strong product instincts and can identify the right problems to solve
  • Are proficient with modern web technologies (React, TypeScript, Python, etc.)
  • Have a track record of building user-facing applications that are actually used and loved by their target audience
  • Communicate effectively with both technical and non-technical stakeholders
  • Care about the societal impacts of your work and are motivated by Anthropic's mission

Preferred qualifications

  • Experience building research tools, scientific software, or experimentation platforms
  • Background in machine learning, AI research, or working closely with ML researchers
  • Founded or been an early engineer at a startup, particularly one focused on developer or researcher tools
  • Built open-source tools or platforms with active user communities
  • Experience with data visualization, interactive interfaces, or novel interaction paradigms
  • Contributed to engineering platforms or internal tooling at scale (similar to Heroku, Vercel, or other platform-as-a-service products)
  • Experience leveraging AI/LLMs to build more powerful or efficient tools
  • Previous work in creative tools, artist tools, or other domains requiring deep user empathy
  • Domain knowledge in areas like human-computer interaction, systems safety, or AI alignment

Benefits

  • $300,000—$405,000 USD

Great fits start here.

Whether you're looking for your next role or your next team member, give people and possibilities a place to meet.