Remote Life Sciences Task Architect for AI Evaluation

OpenTrain AI, Inc.; United States; Remote; USD 51,000 - 76,000; Part time

Job summary

OpenTrain AI, Inc. seeks a Computational Life Sciences Task Designer to craft multi-step workflows that test AI agents on data inspection, code execution, problem solving, and result explanation. This remote, part-time contractor role centers on turning real workflows into self-contained tasks for reproducible evaluation.

The ideal candidate holds a Ph.D. or equivalent experience in life sciences, with strong Python programming and a track record of designing gradable, authentic tasks for

Qualifications

• Ph.D., postdoctoral experience, or equivalent research experience in life sciences.
• Strong Python programming experience.
• Experience with multi-step computational analyses and reproducibility.
• Experience designing graded tasks from authentic workflows.

Responsibilities

• Turn real computational life sciences workflows into self-contained tasks for evaluating AI agents.
• Create input files, datasets, instructions, constraints, and expected deliverables.
• Build tasks with executable code, reproducible solutions, and objectively verifiable results.
• Create automated grading criteria to measure reasoning and execution.
• Make sure tasks run reproducibly in controlled environments without network access.
• Use reviewer feedback to maintain task quality and throughput.

Skills

Python programming

Scientific programming

Data inspection

Education

Ph.D. in life sciences

Postdoctoral experience

Tools

Command-line tools

Reproducible pipelines

Job description

OpenTrain AI, Inc. seeks a Computational Life Sciences Task Designer to craft multi-step workflows that test AI agents on data inspection, code execution, problem solving, and result explanation. This remote, part-time contractor role centers on turning real workflows into self-contained tasks for reproducible evaluation.

The ideal candidate holds a Ph.D. or equivalent experience in life sciences, with strong Python programming and a track record of designing gradable, authentic tasks for