ML/AI Systems Engineer

You will design and deploy multi-step AI agents for advanced mathematical and scientific workflows. The work includes turning proofs and dense technical material into structured specifications, automated pipelines, and dependable software systems.

You will work with mathematical researchers to create reusable systems for STEM-focused AI agents and improve how they reason, use tools, manage state, and record execution steps.

• Design workflows for multi-step reasoning, hypothesis generation, and technical validation.
• Build reusable agent architecture, tools, and memory systems for STEM applications.
• Use Python, language model orchestration, structured outputs, tool integrations, and state management.
• Evaluate and improve agent accuracy, reliability, and execution traces.
• Translate complex academic material into modular software systems.

What it pays and takes

This is a remote contractor assignment for candidates across Latin America. The contract lasts one month and requires an expert level of experience with AI agent systems and software engineering.

• Pay: $25 per approved task.
• Time: 20+ hours per week, including four hours of overlap with Pacific Time.
• Location: Open to candidates in the listed Latin American countries.
• Language: Fluent English required.
• Experience: Proven record building and deploying production-grade, multi-step agent systems.
• Technical skills: Advanced Python, language model orchestration, structured output generation, tool integration, and state management.
• Preferred background: AI or machine learning applications in mathematics, physics, computer science, quantitative finance, or another STEM field.
• Communication: Comfortable working with mathematical subject-matter experts and translating dense academic material into software.

How it works

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

About AI training work

AI training is the human work behind systems that learn from examples, including building software, reviewing model outputs, and evaluating how well AI follows complex instructions. People with strong technical or subject-matter expertise help make these systems more accurate and reliable.