Turn scientific expertise into smarter AI
Join a remote STEM project. Design research-level challenges that train next-gen AI. Earn up to $6,000+/month.
Turn scientific expertise into smarter AI
Join a remote STEM project. Design research-level challenges that train next-gen AI. Earn up to $6,000+/month.
Turn scientific expertise into smarter AI
Join a remote STEM project. Design research-level challenges that train next-gen AI. Earn up to $6,000+/month.
About the project
About the project
Professionals in various STEM domains contribute their expertise to help advance AI systems that reason through complex scientific problems. Experts on this project design challenges, build reasoning chains, and verify outputs using Python.
Project details
Up to $6,000+/month
Flexible remote project
Research-level AI tasks
Performance bonus up to 20%
How you might contribute
How you might contribute
Mathematics
Design a problem requiring the AI to compute sharp constants in Fourier approximation bounds, then verify convergence through numerical optimization in Python.
Biology
Construct a DNA sequence alignment challenge with iterative transformations and affine-gap scoring, then confirm the AI's dynamic programming solution is correct.
Engineering
Create a battery degradation simulation coupling thermal feedback, capacity fade, and state-of-health estimation, then verify each step computationally.
Physics
Build a multi-step challenge around bound-state energies in a quantum well with discontinuous parameters, and validate boundary conditions and eigenvalue accuracy.
Computer Science
Formulate a distributed query estimation problem across partitioned databases and check whether the AI correctly handles shard alignment, filters, and join logic.
Mathematics
Design a problem requiring the AI to compute sharp constants in Fourier approximation bounds, then verify convergence through numerical optimization in Python.
Biology
Construct a DNA sequence alignment challenge with iterative transformations and affine-gap scoring, then confirm the AI's dynamic programming solution is correct.
Engineering
Create a battery degradation simulation coupling thermal feedback, capacity fade, and state-of-health estimation, then verify each step computationally.
Physics
Build a multi-step challenge around bound-state energies in a quantum well with discontinuous parameters, and validate boundary conditions and eigenvalue accuracy.
Computer Science
Formulate a distributed query estimation problem across partitioned databases and check whether the AI correctly handles shard alignment, filters, and join logic.
Mathematics
Design a problem requiring the AI to compute sharp constants in Fourier approximation bounds, then verify convergence through numerical optimization in Python.
Biology
Construct a DNA sequence alignment challenge with iterative transformations and affine-gap scoring, then confirm the AI's dynamic programming solution is correct.
Engineering
Create a battery degradation simulation coupling thermal feedback, capacity fade, and state-of-health estimation, then verify each step computationally.
Physics
Build a multi-step challenge around bound-state energies in a quantum well with discontinuous parameters, and validate boundary conditions and eigenvalue accuracy.
Computer Science
Formulate a distributed query estimation problem across partitioned databases and check whether the AI correctly handles shard alignment, filters, and join logic.
Mathematics
Design a problem requiring the AI to compute sharp constants in Fourier approximation bounds, then verify convergence through numerical optimization in Python.
Biology
Construct a DNA sequence alignment challenge with iterative transformations and affine-gap scoring, then confirm the AI's dynamic programming solution is correct.
Engineering
Create a battery degradation simulation coupling thermal feedback, capacity fade, and state-of-health estimation, then verify each step computationally.
Physics
Build a multi-step challenge around bound-state energies in a quantum well with discontinuous parameters, and validate boundary conditions and eigenvalue accuracy.
Computer Science
Formulate a distributed query estimation problem across partitioned databases and check whether the AI correctly handles shard alignment, filters, and join logic.
Mathematics
Design a problem requiring the AI to compute sharp constants in Fourier approximation bounds, then verify convergence through numerical optimization in Python.
Biology
Construct a DNA sequence alignment challenge with iterative transformations and affine-gap scoring, then confirm the AI's dynamic programming solution is correct.
Engineering
Create a battery degradation simulation coupling thermal feedback, capacity fade, and state-of-health estimation, then verify each step computationally.
Physics
Build a multi-step challenge around bound-state energies in a quantum well with discontinuous parameters, and validate boundary conditions and eigenvalue accuracy.
Computer Science
Formulate a distributed query estimation problem across partitioned databases and check whether the AI correctly handles shard alignment, filters, and join logic.
Mathematics
Design a problem requiring the AI to compute sharp constants in Fourier approximation bounds, then verify convergence through numerical optimization in Python.
Biology
Construct a DNA sequence alignment challenge with iterative transformations and affine-gap scoring, then confirm the AI's dynamic programming solution is correct.
Engineering
Create a battery degradation simulation coupling thermal feedback, capacity fade, and state-of-health estimation, then verify each step computationally.
Physics
Build a multi-step challenge around bound-state energies in a quantum well with discontinuous parameters, and validate boundary conditions and eigenvalue accuracy.
Computer Science
Formulate a distributed query estimation problem across partitioned databases and check whether the AI correctly handles shard alignment, filters, and join logic.
Mathematics
Design a problem requiring the AI to compute sharp constants in Fourier approximation bounds, then verify convergence through numerical optimization in Python.
Biology
Construct a DNA sequence alignment challenge with iterative transformations and affine-gap scoring, then confirm the AI's dynamic programming solution is correct.
Engineering
Create a battery degradation simulation coupling thermal feedback, capacity fade, and state-of-health estimation, then verify each step computationally.
Physics
Build a multi-step challenge around bound-state energies in a quantum well with discontinuous parameters, and validate boundary conditions and eigenvalue accuracy.
Computer Science
Formulate a distributed query estimation problem across partitioned databases and check whether the AI correctly handles shard alignment, filters, and join logic.
Mathematics
Design a problem requiring the AI to compute sharp constants in Fourier approximation bounds, then verify convergence through numerical optimization in Python.
Biology
Construct a DNA sequence alignment challenge with iterative transformations and affine-gap scoring, then confirm the AI's dynamic programming solution is correct.
Engineering
Create a battery degradation simulation coupling thermal feedback, capacity fade, and state-of-health estimation, then verify each step computationally.
Physics
Build a multi-step challenge around bound-state energies in a quantum well with discontinuous parameters, and validate boundary conditions and eigenvalue accuracy.
Computer Science
Formulate a distributed query estimation problem across partitioned databases and check whether the AI correctly handles shard alignment, filters, and join logic.
Mathematics
Design a problem requiring the AI to compute sharp constants in Fourier approximation bounds, then verify convergence through numerical optimization in Python.
Biology
Construct a DNA sequence alignment challenge with iterative transformations and affine-gap scoring, then confirm the AI's dynamic programming solution is correct.
Engineering
Create a battery degradation simulation coupling thermal feedback, capacity fade, and state-of-health estimation, then verify each step computationally.
Physics
Build a multi-step challenge around bound-state energies in a quantum well with discontinuous parameters, and validate boundary conditions and eigenvalue accuracy.
Computer Science
Formulate a distributed query estimation problem across partitioned databases and check whether the AI correctly handles shard alignment, filters, and join logic.
Mathematics
Design a problem requiring the AI to compute sharp constants in Fourier approximation bounds, then verify convergence through numerical optimization in Python.
Biology
Construct a DNA sequence alignment challenge with iterative transformations and affine-gap scoring, then confirm the AI's dynamic programming solution is correct.
Engineering
Create a battery degradation simulation coupling thermal feedback, capacity fade, and state-of-health estimation, then verify each step computationally.
Physics
Build a multi-step challenge around bound-state energies in a quantum well with discontinuous parameters, and validate boundary conditions and eigenvalue accuracy.
Computer Science
Formulate a distributed query estimation problem across partitioned databases and check whether the AI correctly handles shard alignment, filters, and join logic.
Mathematics
Design a problem requiring the AI to compute sharp constants in Fourier approximation bounds, then verify convergence through numerical optimization in Python.
Biology
Construct a DNA sequence alignment challenge with iterative transformations and affine-gap scoring, then confirm the AI's dynamic programming solution is correct.
Engineering
Create a battery degradation simulation coupling thermal feedback, capacity fade, and state-of-health estimation, then verify each step computationally.
Physics
Build a multi-step challenge around bound-state energies in a quantum well with discontinuous parameters, and validate boundary conditions and eigenvalue accuracy.
Computer Science
Formulate a distributed query estimation problem across partitioned databases and check whether the AI correctly handles shard alignment, filters, and join logic.
Mathematics
Design a problem requiring the AI to compute sharp constants in Fourier approximation bounds, then verify convergence through numerical optimization in Python.
Biology
Construct a DNA sequence alignment challenge with iterative transformations and affine-gap scoring, then confirm the AI's dynamic programming solution is correct.
Engineering
Create a battery degradation simulation coupling thermal feedback, capacity fade, and state-of-health estimation, then verify each step computationally.
Physics
Build a multi-step challenge around bound-state energies in a quantum well with discontinuous parameters, and validate boundary conditions and eigenvalue accuracy.
Computer Science
Formulate a distributed query estimation problem across partitioned databases and check whether the AI correctly handles shard alignment, filters, and join logic.
Types of STEM AI tasks
Tasks vary between projects, but generally center around prompting AI with writing challenges, evaluating its output, and improving the results until they meet professional standards.
Tasks vary between projects, but generally center around prompting AI with writing challenges, evaluating its output, and improving the results until they meet professional standards.
Create prompts
Create challenging problems across various STEM domains (science, technology, engineering, mathematics) that push AI to the limits of scientific reasoning.
Evaluate AI outputs
Review model-generated answers line by line. Identify misapplied formulas, confused principles, and incorrect boundary conditions.
Write the ideal version
Author a complete, rigorous solution that demonstrates the correct reasoning path to form a training signal for the model’s next generation.
Create prompts
Create challenging problems across various STEM domains (science, technology, engineering, mathematics) that push AI to the limits of scientific reasoning.
Evaluate AI outputs
Review model-generated answers line by line. Identify misapplied formulas, confused principles, and incorrect boundary conditions.
Write the ideal version
Author a complete, rigorous solution that demonstrates the correct reasoning path to form a training signal for the model’s next generation.
Sample fictional tasks for expert evaluation.
Figures and parameters are illustrative and designed to test scientific reasoning.
Sample fictional tasks for expert evaluation.
Figures and parameters are illustrative and designed to test scientific reasoning.
Earning potential
Project duration is ~2 months. All numbers are approximated.
Final pay depends on completed tasks.
Up to
$3,000/mo
$3,000/mo
Up to
$6,000/mo
$6,000/mo
Up to
$12,000/mo
$12,000/mo
10 hours/week
80 hours in total
10 hours per week
20 hours/week
160 hours in total
20 hours per week
35 hours/week
320 hours in total
35 hours per week
Referral program
Get up to $300 for every STEM expert you bring in
Performance bonus
Up to 20% extra pay. Based on 10hr min weekly participation, quality and complexity
Current opportunities
How it works
1
Apply
Indicate your domain and experience
2
Qualify
Complete a domain-specific assessment
3
Onboard
Get access to the platform
Get access to the platform
4
Earn
Complete tasks on your own schedule
Frequently
asked questions
Frequently asked questions
What STEM domains do you need?
Mathematics, Physics, Biology, Chemistry, Data Science, ML, Statistics, and Engineering disciplines. Math and Physics are the highest priority.
What will I do?
Design research-level challenges, build reasoning paths, verify solutions in Python, and review AI outputs.
How much can I earn?
$3,000–$12,000+ per project depending on hours, plus up to 20% performance bonus.
Do I need a PhD?
Preferred, not required. We need research-level depth from any background.
Do I need Python?
Yes. You should be comfortable writing verification scripts.
Is this freelance?
Yes — project-based, no employment contract, no minimum hours.
What's Mindrift
What's Mindrift
Mindrift connects domain experts with AI training projects. Contribute remotely, choose your schedule, help build AI that understands science.
