STEM Researcher for AI Model Training [Remote]

Apply your STEM domain expertise to review, validate, and document agentic AI session traces used in technical research. Your analyses and written evaluations will help train next generation AI systems by identifying real-world problem solving, multi-step reasoning, and points of human guidance within agent-assisted workflows.

• Review, curate, and submit historical session traces from local or agentic AI tools that demonstrate substantive STEM research or technical workflows.
• Analyze and validate the technical depth and integrity of AI-assisted problem solving, confirming sessions include multi-step reasoning and authentic human intervention.
• Document research methodologies, step-by-step problem solving approaches, and explicit points where humans guided or corrected the AI.
• Evaluate strengths and limitations of agentic AI tools in STEM research contexts, and highlight areas for model training or improvement.
• Produce clear, concise written explanations describing session processes, outputs, and validation steps.
• Provide constructive, discipline-appropriate feedback on AI performance, accuracy, and reasoning based on professional STEM experience.

• Advanced academic or professional background in a STEM field, such as science, technology, engineering, or mathematics.
• Strong scientific reasoning, technical research, data analysis, and code debugging skills.
• Experience documenting research methods, producing technical writing, and validating outputs in complex problem solving scenarios.
• Familiarity with AI evaluation, output validation, and iterative research methods is highly valued.
• Preferred first-hand experience using local or agentic AI tools for research or technical projects, beyond simple web-based chatbots.
• Access to and familiarity with historical session data that capture STEM or technical workflows is required.
• Excellent written and verbal communication skills to clearly articulate technical findings and feedback.

• Engagement type, independent contractor.
• Remote work, location flexible.
• Work focuses on reviewing historical session traces from agentic or local AI tools, including but not limited to Claude Code, Claude Cowork, Codex, Claude for Life Sciences, and OpenCode.

Pay range: 80 to 100 USD per hour.

• No prior professional experience in AI is required, domain expertise in STEM is the primary qualification.
• Candidates must have access to historical session traces or similar data from agentic or local AI tools that demonstrate technical workflows.
• This role is open to remote contractors; any additional work authorization or local contracting requirements are the candidate''s responsibility.

Candidates will be identified and vetted through the company''s talent platform and recruiter system. Successful applicants will be asked to demonstrate relevant STEM experience and familiarity with session trace data as part of the vetting and onboarding process.