Baseten
Software Engineer - AI Developer Productivity
About this role
Build Baseten's internal AI developer platform that makes AI-native software development the default for engineers. Own agent infrastructure, context layers, evaluation systems, and adoption mechanics that help teams ship AI-assisted features safely and effectively.
What you'll do
- Design and operate end-to-end internal AI developer platform including architecture, deployment, and measurement
- Integrate third-party AI tools (Claude Code, Cursor, etc.) with context layers that work against the monorepo
- Build frameworks enabling engineers to create custom agents without deep LLM expertise
- Establish evaluation practices to measure AI-assisted development impact on cycle time and quality
- Drive adoption through superior defaults and developer experience rather than mandates
- Implement safety layer including permissions, secrets handling, audit trails, and cost controls
What they're looking for
- Python and/or Go for building developer tools
- LLM frameworks, tool calling, MCP, and agentic orchestration patterns
- Agent design and failure handling in production
- CI/CD, Kubernetes, Docker fundamentals
- Internal developer platform design and adoption strategy
- AI coding tools (Claude Code, Cursor) with deep understanding of limitations
- Evaluation and metrics design for AI-assisted workflows
- Monorepo context management and tooling
Opens the application — the Jobs AI extension fills it for you. Set up autofill
Opens the official application on the employer’s site. No login required.
Baseten
Baseten builds an AI inference platform that enables companies to deploy and manage machine learning models in production at scale. The company is hiring software engineers, infrastructure specialists, and product engineers to develop observability systems, frontend experiences, reliability infrastructure, developer tools, and enterprise deployment solutions.
- Website
- baseten.com
Likely interview questions
- Walk us through an agentic system you shipped to real users—what patterns worked and where did it break?
- How would you approach building context infrastructure that keeps agents competent as a monorepo evolves?