🔬 Physical Sciences Expert – AI Training | Remote
🔬 Physical Sciences Expert – AI Training | Remote
Are you a physicist, chemist, materials scientist, or physical-science researcher with strong programming skills? This short-term AI Training opportunity could be a great chance to use your scientific expertise to help improve the next generation of AI systems.
Braintrust is looking for Physical Sciences Experts who can create challenging, reproducible scientific tasks that AI agents must solve in a Linux terminal.
💼 Job Overview
Position: Physical Sciences Expert – AI Training
Job Type: Freelance
Category: Science & Research
Work Mode: Remote
Locations: Bangladesh, Brazil + 7 other locations
Rate: 💰 $300 per approved task
Experience: 1–3 years
Project Length: Approximately 5 weeks
Availability: 40 hours/week
Timezone: PST/PDT/CIST – partial overlap
Top Skill: Experimental Physics
Other Skill: Python
⚠️ Important: Applicants should review the timezone requirements before applying.
🧪 About the Project
The project focuses on evaluating how AI systems handle real scientific problems.
You will transform physics, chemistry, materials science, and related research workflows into challenging tasks that an AI agent must complete using a Linux terminal environment.
The goal is to create scientific tasks that are realistic, measurable, reproducible, and technically challenging.
🔬 What Will You Do?
As a Physical Sciences Expert, you may:
Design multi-step scientific tasks
Create simulations and scientific models
Develop quantitative analysis workflows
Build computational experiments
Prepare scientific inputs and working environments
Create automated tests to evaluate AI-generated solutions
Debug scientific calculations and dependencies
Document units and assumptions
Define numerical tolerances
Identify scientific limitations
Validate that workflows produce accurate and reproducible results
🧠 What Are They Looking For?
Candidates should have:
A PhD, postdoctoral experience, or equivalent advanced technical experience in a relevant physical-science discipline
Strong programming skills in Python, C/C++, Julia, Bash, or another scientific programming language
Experience working in Linux or terminal environments
Experience independently creating and validating scientific workflows
Understanding of numerical precision and physical constraints
Knowledge of reproducible scientific research
Experience with benchmark design, research software, publications, or open-source projects is welcome but not mandatory.
💻 Ideal Background
This opportunity may be suitable for professionals with backgrounds in:
Physics
Experimental Physics
Chemistry
Materials Science
Computational Science
Scientific Research
Computational Physics
Scientific Programming
⏰ Project Details
📅 Expected duration: 5 weeks
⏱️ Availability: 40 hours/week
🌎 Timezone: Partial overlap with Pacific working hours
💻 Work: Remote
The project requires approximately 4 hours of overlap with Pacific working hours each day.
💰 Compensation
The listed compensation is:
$300 per approved task
Each task is estimated to take approximately 6–8 hours on average.
The actual total earnings will depend on the number of tasks completed and approved.
📝 Selection Process
The selection process includes:
Short interest form
60-minute assessment
Final review with the team
🚀 Why Apply?
This is an interesting opportunity for scientific professionals who want to combine their research expertise, programming skills, and AI to help evaluate and improve advanced AI systems.
If you have a strong physical-science background and enjoy computational research and scientific problem-solving, this project could be worth exploring.
httpApply Now