Project Based Recruitment | Scientific Coding - Materials Science & Python | Remote

About the role

A new project‑based opportunity for Materials Science and Python experts is available for professionals interested in contributing to advanced AI training and evaluation. The project focuses on developing rigorous scientific coding tasks for frontier AI models, combining materials science knowledge with scientific programming, problem design, Python implementation, and AI evaluation. Selected contributors will create complex scientific problems, develop verified Python solutions, build comprehensive unit tests, and design test cases that can distinguish accurate AI‑generated solutions from incorrect ones. This is an intensive 8‑week contractor project requiring up to 40 hours per week and a minimum of four hours of overlap with PST. Candidates from Indonesia and several other eligible countries can apply. For professionals exploring remote wfa jobs, this project offers a particularly relevant opportunity to combine academic STEM expertise with AI model training. The work involves transforming complex STEM concepts into structured programming challenges that AI models must solve, ensuring each task is scientifically accurate, logically structured, deterministic, testable, and sufficiently challenging to distinguish strong model reasoning from incorrect or incomplete solutions.

Responsibilities

• Write scientific problem specifications consisting of one main problem and at least three connected sub‑problems.
• Design sub‑problems that progressively build toward the main solution.
• Implement verified golden solutions using Python.
• Provide complete unit test coverage.
• Create discriminative test cases that distinguish correct and incorrect model outputs.
• Validate scientific correctness and computational logic.
• Run quality‑control checks through the task platform.
• Complete structural validation and quality‑rubric checks.
• Improve tasks based on quality‑control feedback.
• Optimize tasks for Pass@K evaluation across multiple AI model judges.
• Maintain high‑quality submissions with minimal rework.
• Participate in project reviews, feedback sessions, and standups during the required overlap hours.

Requirements

• Remote contractor/freelancer arrangement.
• 8‑week contract duration.
• Work commitment: 40 hours per week.
• Time‑zone requirement: minimum 4 hours of overlap with PST.
• No medical or paid leave under the contractor arrangement.
• Eligible locations: Bangladesh, Brazil, Colombia, Egypt, Ghana, India, Pakistan, Indonesia, Kenya, Nigeria, Turkey, Vietnam.

Qualifications

• Master’s or PhD degree in Materials Science or a closely related discipline.
• Strong Python programming skills.
• Experience with scientific computing.
• Ability to formulate rigorous scientific problems.
• Ability to define clear constraints and expected outputs.
• Strong attention to detail.
• Understanding of scientific correctness and determinism.
• Experience with AI data annotation, research, or scientific writing.
• Familiarity with LLM evaluation frameworks or coding benchmarks.
• Experience with scientific Python libraries such as NumPy, SciPy, or SymPy.
• Published research or academic project experience in a STEM field.

Skills

• Materials science and applied chemistry expertise.
• Proficiency with Python and scientific libraries (NumPy, SciPy, SymPy).
• Ability to translate scientific concepts into computational problems.
• Experience designing discriminative test cases for AI evaluation.
• Knowledge of LLM evaluation and AI benchmark development.
• Background in scientific writing, data annotation, or coding benchmarks.

Schedule

40