Research Scientist, Part-Time AI Research Evaluation - Remote (United States)

OpenTrain AI; États-Unis; À distance; USD 138 000 - 551 000; Temps partiel

Avantages offerts par ce poste

Fully remote

Flexible schedule

Independent contractor

Résumé du poste

OpenTrain AI seeks experienced researchers to evaluate AI-generated research outputs, focusing on proofs, algorithms, and simulations. You will examine correctness, originality and provide clear written explanations of your reasoning.

The role is fully remote with a flexible, independent schedule for part-time contractors in the United States. A strong publication record and practical skills in numerical methods and AI evaluation are valued.

Qualifications

• Strong publication record in computer science, computational physics or mathematical physics.
• Experience evaluating AI-generated proofs, algorithms and simulations.
• Ability to clearly explain reasoning and assessment to researchers.

Responsabilités

• Define research questions and focus areas for evaluation.
• Test proofs, algorithms and simulations with AI models and agents.
• Document findings, limitations and directions for future work.
• Review relevant literature to assess novelty and soundness and report approaches.

Connaissances

Research judgement

Numerical methods

Algorithm implementation

English fluency

Formation

PhD candidate or equivalent

Postdoctoral researcher

Outils

Lean

SAT/SMT solvers

Description du poste

Evaluate computer science and computational physics research produced by AI models, on your own schedule.

Research outputs from AI systems need careful technical review. This role brings together researchers who can examine a proposed proof, question an algorithm's reasoning, or check whether a simulation supports its conclusions. The work involves testing ideas and recording what the evidence shows, including incomplete results and limitations. You will assess both correctness and originality, with written explanations that make your reasoning clear to other researchers.

What you'll do

• Choose research questions in your specialty, including unresolved problems and possible improvements to known bounds
• Explore proofs, algorithms and computational simulations with AI models and agents
• Scrutinize proof attempts and numerical outputs, drawing on formal methods where useful
• Read relevant publications to check novelty and soundness, then report your approach, partial findings, difficulties and possible next directions

Why this is worth your time

• Get paid $100–$400 an hour to evaluate AI-generated research outputs
• Apply your research methods to questions involving proofs, algorithms and simulations
• Examine where an AI-generated result succeeds, where it fails and what needs further investigation
• Fully remote and flexible, no fixed shifts, no commute

Who this is for

• Late-stage PhD candidates, postdoctoral researchers or early-career academics up to five years after the PhD
• Advanced research ability in computer science, computational physics or mathematical physics
• A strong recent publication record, with research addressing unresolved questions or introducing new concepts
• Practical skills in simulation, numerical methods and implementing algorithms
• Careful assessment of proofs, numerical findings, algorithms and scientific claims
• Professional fluency in English
• Helpful experience includes recognition for a paper or dissertation, reviewing conference submissions, research using AI models or agents, or formal methods such as Lean and SAT/SMT solvers
• A public research statement or homepage explaining the open questions you study is welcome
• No AI experience needed.

Pay and details

• $100–$400 per hour, paid hourly as an independent contractor.
• Part-time, flexible schedule; fully remote, United States

Ready to put your research judgment to work evaluating AI systems.

About OpenTrain AI

OpenTrain AI connects experienced professionals with paid AI training work. This role involves reviewing AI-generated research and assessing the technical reasoning behind it.