We are running a paid project focused on creating and validating complex coding tasks for AI systems. The goal is to build realistic, self-contained software engineering problems that push the boundaries of automated coding agents. Your contributions will directly improve how these systems navigate real-world development environments.
How It WorksYou will design intricate software problems that an AI agent must solve within a remote terminal environment. This involves implementing features, debugging codebases, configuring Docker containers, and writing deterministic automated tests. You will also create reference solutions to ensure every task can be graded reliably and accurately. Participants work independently to verify that each problem is challenging, solvable, and completely self-contained.
Who This Is ForWe welcome experienced software engineers who are highly comfortable working independently in unfamiliar codebases. You should have strong command-line skills, proficiency with GitHub, and experience writing clean tests. Candidates with a background in complex debugging, system configuration, or automated testing are highly encouraged to apply.
What You'll DoDesign realistic, self-contained software engineering problems for AI evaluation.
Configure development environments and work within Docker containers.
Write robust, deterministic automated tests to evaluate task completion.
Create verified reference solutions to ensure tasks are accurately graded.
Professional experience working as a software engineer.
Strong proficiency with command-line tools and GitHub workflows.
Comfortable navigating and debugging unfamiliar codebases independently.
Willingness to complete an identity verification and skills assessment.
$82 per task
Ready to participate?
Start your paid interview now
About Terac
Terac is building the world's largest pool of vetted human experts for AI. Researchers, AI labs, and product teams use Terac to recruit, screen, and pay study participants across industries, languages, and skill sets.
Learn more at terac.com or on YouTube at @jointerac.
.jpeg)
