Senior Software Engineer — Agentic Coding

Autonomous coding evaluation

Build, test, and improve systems in which AI plans software work, writes code, runs tests, and iterates on results. This opportunity is aimed at senior engineers in Seattle and focuses on practical software delivery with limited human intervention.

• Design and refine workflows for autonomous coding agents.
• Review generated code for correctness, efficiency, security, and sound engineering practices.
• Create realistic programming challenges and benchmarks that stress-test agent capabilities.
• Give expert feedback on reasoning sequences, tool selection, and output quality.
• Identify failure patterns, unusual inputs, and situations where agents cannot complete their plans reliably.
• Develop evaluation frameworks, scoring criteria, and testing harnesses.
• Collaborate asynchronously with researchers and engineers working on LLMs and AI agents.

Required experience

At least five years of professional software engineering experience is required, including building, delivering, and maintaining real systems. You need proficiency in one or more major languages such as Python, JavaScript or TypeScript, Java, C++, Go, or Rust.

Strong knowledge of architecture, design patterns, and system design is essential. You must be comfortable reading, reviewing, and debugging unfamiliar code across varied repositories and coding styles.

Practical understanding of version control, CI/CD, test frameworks, and code review is required. The role also calls for clear reasoning through ambiguous problems with several steps, precise technical explanations, and reliable independent delivery.

Interest in AI, language models, and autonomous software development is important to the work.

Helpful background

AI or ML systems, prompt engineering, and LLM tools are useful experience. Familiarity with LangChain, AutoGPT, CrewAI, or OpenAI function calling is welcome.

Developer tools, compilers, static analysis, code generation, RLHF, preference labeling, and other model evaluation or training work are relevant. Open-source contributions, a strong public engineering portfolio, and knowledge of multiple languages and programming paradigms are also helpful.

Contract terms

The fully remote hourly engagement offers flexible independent assignments for 10 to 40 hours weekly. Work involves international collaboration with AI researchers and ongoing exposure to language models and autonomous systems.

Further projects may support longer engagements or extensions as the scope grows.