Agent Engineer

About the role

ABOUT COFRAME

At Coframe, we are building a future where user and agent interfaces can adapt, evolve, and personalize themselves. We are generating revenue, growing rapidly, and trusted by companies you definitely know.

Coframe is a primarily in-person company based in San Francisco, well-capitalized with backing from Khosla Ventures, Nat Friedman, the founder of Android, and others. Our team is small and talent-dense, consisting of competitive programmers, former founders (including our founder who previously co-founded a unicorn), and AI researchers from top institutions who have created some of the most popular open-source code generation projects.

ABOUT THE ROLE:

As an Agent Engineer, you’ll build the core intelligence behind these systems. You’ll work on AI agents that create production-ready interfaces, reason through complex tasks, use tools, validate their own work, and continuously improve through real-world feedback.

Each new capability expands the range of experiences Coframe can create and optimize. Your work will increase experimentation velocity, improve output quality, and reduce the human effort required to build and maintain digital experiences.

YOU’RE A GREAT FIT IF YOU HAVE:

- Strong software engineering fundamentals and experience shipping reliable production systems

- Experience building LLM applications, AI agents, autonomous workflows, or code generation systems

- Experience with prompting, evaluation, retrieval-augmented generation, tool use, or agent frameworks

- Experience building systems for planning, reasoning, and multi-step execution

- The ability to design reliable systems around nondeterministic model behavior

- Experience with automated testing, evaluation systems, and parallel execution

- Strong web engineering skills and product judgment for what makes a high-quality interface

- Experience integrating agents with browsers, development tools, APIs, or internal platforms

- Comfort working across the stack to solve complex engineering problems

- The ability to turn qualitative feedback into measurable improvements to agent performance

- Curiosity, pragmatism, and excitement about working on problems with no established playbook

- A track record of exploring new approaches to code generation and agent development beyond standard tools and frameworks

WHAT YOU’LL DO:

- Build AI agents that generate high-quality user interfaces, client-side features, and product experiences

- Develop agent workflows for planning, reasoning, tool use, and multi-step execution

- Build systems that allow agents to autonomously generate, test, verify, and improve their own work

- Create validation systems that catch bugs, off-brand outputs, and poor user experiences before launch

- Develop online and offline evaluation systems for generated code, designs, and user experiences