Software Engineer, AI Research and Evaluation (US-Based)

Create and review coding examples used to train and evaluate AI models. The work spans systems programming, infrastructure, backend development, and performance-focused software.

• Create precise coding examples and solutions in Python, C, C++, Rust, Go, Java, and JavaScript, including ReactJS.
• Review and correct AI-generated code for correctness, performance, scalability, and reliability.
• Assess AI capabilities in prototyping, architecture and API design, production development, monitoring, and ongoing operations.
• Build agents to check systems and infrastructure code, find recurring errors, and create automated ways to verify software solutions.
• Work with cross-functional teams to improve AI coding solutions against performance benchmarks.

What it pays and takes

• Pay: Not specified.
• Time: 20 or more hours per week.
• Location: Open to candidates in the United States.
• Work type: Contract and part-time.
• Experience: At least three years of software engineering experience.
• Expertise: Strong skills in systems programming, infrastructure, or backend development, plus experience building and deploying scalable, production-grade software.
• Skills: Deep knowledge of software architecture, design, development, debugging, and code-quality review. You should be able to explain your evaluation clearly in writing and conversation.
• Programming: Proficiency in relevant languages such as Python, C or C++, Rust, Go, Java, or JavaScript.
• Language: Fluent in English.
• Helpful background: Experience with advanced intelligent systems, frontier AI, or comparable production software environments. Strong technical ability and relevant engineering experience are valued alongside formal education.

How it works

Apply on OpenTrain with your resume, then complete the application on the hiring site.

About AI training work

AI training work uses human-created examples and reviews to help AI models learn. Software engineers are needed to assess whether generated code is correct, efficient, scalable, and reliable.