Software Engineer - AI Expert

Description

The company builds the intelligence behind frontier AI and needs experienced software engineers for paid, project-based work that helps train and evaluate the next generation of AI coding systems.

You work remotely and asynchronously, on projects that fit around your current schedule.

Your engineering experience shows AI systems what good software engineering looks like.

More information is available on the company website.

Responsibilities

• Depending on the project, you may build, evaluate, debug, review, demonstrate, create evaluations, or compare solutions.
• Build: Solve realistic engineering problems and produce high-quality implementations.
• Evaluate: Review AI-generated code, pull requests, architectures, tests, and debugging approaches.
• Debug: Find the cause of failures in real or realistic repositories and explain what went wrong.
• Review: Judge whether an AI agent's implementation is correct, maintainable, secure, and ready for production.
• Demonstrate: Work through full engineering workflows in real development environments so AI systems can learn how experienced engineers work.
• Create evaluations: Write challenging engineering tasks and criteria for measuring AI coding agents.
• Compare solutions: Assess several AI-generated solutions and identify meaningful differences in correctness, quality, and judgment.
• Projects may cover repository navigation, implementation, testing, debugging, code review, system design, and other engineering workflows.

Requirements

• You are an experienced working software engineer. AI research experience is not necessary.
• Required: 3+ years of professional software engineering experience.
• Required: Strong programming and debugging skills.
• Required: Experience working in production codebases.
• Required: Ability to understand unfamiliar repositories and systems.
• Required: Strong code review and technical judgment.
• Required: Experience writing and reviewing tests.
• Required: Ability to clearly explain why a solution is correct or incorrect.
• Valuable, but not required: experience with Python, JavaScript/TypeScript, Go, Java, C++, Rust, or similar languages.
• Valuable, but not required: backend or distributed systems, APIs and microservices, databases and data systems, cloud infrastructure, DevOps or CI/CD, security, frontend engineering, mobile development, machine learning infrastructure, or agentic AI and coding agents.
• You do not need prior AI experience. Knowing how to do software engineering well matters most.
• Example tasks: fix a bug in an unfamiliar repository; review an AI-generated pull request and list its issues; implement a feature with suitable tests; compare two AI implementations and decide which one works; debug a failing application and explain the root cause; judge whether an AI coding agent finished an engineering task; complete a multi-step task in a real development environment.
• The aim is not a quick check of whether code looks right, but to capture how an experienced engineer decides that software actually works.

Conditions

Paid, project-based work with no fixed schedule and no minimum commitment.

Rates depend on your experience, specialization, and project requirements. Your rate and project scope are shown before you commit.

Some projects are short. Others may recur over a longer period.

Remote, asynchronous, and flexible in hours. You can fit it around an existing job or other commitments.

Pay applies to qualifying project work only.

A plausible-looking patch is not the same as sound engineering. AI needs to learn how experienced engineers understand unfamiliar systems, make architectural tradeoffs, debug failures, write reliable tests, review code, handle edge cases, reason about security and performance, and decide when work is production-ready.

Your judgment helps shape how AI learns to build software.

Application process: submit your engineering background, experience, and areas of expertise; complete a short, paid assessment matched to the work; once your experience fits an active project, the scope, requirements, and pay are shared; complete the project remotely and asynchronously; receive the agreed rate for completed work.

Benefits

• You are paid for completed project work at the agreed rate.