The work
You will design and deploy multi-step agent workflows used to evaluate large language models and build training datasets. The work focuses on realistic software engineering tasks and advanced mathematical reasoning problems.
You will connect mathematical research with practical software systems by turning proofs and scientific workflows into structured, testable pipelines.
- Build workflows for multi-step reasoning, hypothesis generation, and technical validation.
- Turn abstract mathematical proofs into technical specifications and automated pipelines.
- Create reusable tools, frameworks, and memory systems for STEM-focused AI agents.
- Evaluate agent performance and improve accuracy, reliability, and traceability across execution steps.
- Help develop verifiable software engineering tasks from public repository histories and human-reviewed datasets.
What it pays and takes
This is a remote contractor assignment marked as entry level. The work requires strong technical experience with AI systems, Python, and multi-step agent orchestration.
- Pay: $30 per approved task.
- Contract: One month.
- Schedule: At least 4 hours per day and 40 hours per week, including 4 hours of overlap with Pacific Standard Time.
- Location: Open to contractors in India, Pakistan, Nigeria, Kenya, Egypt, Ghana, Bangladesh, Turkey, and Mexico.
- Language: Fluency in English.
- Experience: Proven experience architecting and deploying production-grade, multi-step agent systems.
- Technical skills: Advanced Python, large language model orchestration, structured output generation, tool integration, and state management.
- Frameworks: Experience with LangGraph, AutoGen, CrewAI, or custom orchestration systems.
- Useful background: Building AI or machine-learning applications in mathematics, physics, computer science, or quantitative finance.
- Communication: Ability to turn dense academic material into modular software systems and work with mathematical researchers.
How it works
You apply on OpenTrain with your resume and then complete the application on the hiring site.
About AI training work
AI training work is the human side of building artificial intelligence. People prepare and review examples, test model outputs, and create software or reasoning tasks that help AI systems become more accurate and reliable.
People with strong technical or subject knowledge are paid for work that requires careful judgment, such as evaluating code, checking mathematical reasoning, and designing reliable model tests.