Math Expert to Create Original Graduate-Level Multimodal Questions
Budget: $10 – $30 USD
We are seeking experienced Math educators and subject matter experts to develop a set of original, high-level question & answer pairs, at least half of which must be multimodal - containing essential visual components. This project supports the evaluation and training of advanced AI systems through the creation of reasoning-intensive, graduate-level problems that are resistant to surface-level AI solutions.
Project Overview
Our goal is to build a challenging and novel dataset that pushes the boundaries of current AI capabilities in question answering. The questions must test deep conceptual understanding, multi-step reasoning, and problem-solving mathematics.
At least 50% of the questions must incorporate visuals (e.g., graphs, diagrams, models) that are essential to solving the problem. Each question will be tested against Gemini 2.5 Pro and must include a conversation log demonstrating the model’s performance with and without hints.
Challenges
- Questions must be challenging for current AI models, requiring expert-level knowledge and reasoning.
- Questions should be "Google-proof", meaning they cannot be easily answered through simple web searches.
- Answers must be unambiguous and verifiable to ensure consistency in AI evaluation.
Responsibilities
- Create high-difficulty questions (~graduate-level or upper undergraduate level) in your area of expertise.
- Write a detailed explanation and/or solution for each question.
- Ensure questions are original and not copied from textbooks or previous tests.
- Must be able to provide concise, informative hints designed to help advanced AI models reason through and answer the question correctly.
Qualifications
We are seeking individuals with the following profile:
- Education: A Master’s degree (or higher) in a STEM discipline is nice to have. Ph.D. holders are strongly preferred.
- Expertise: Deep academic or professional knowledge in mathematics, physics, chemistry, biology, or a related field.
- Writing Skills: Excellent written English; ability to write clear, rigorous questions and explanations.
- Reasoning Skills: Ability to craft problems requiring advanced, multi-step reasoning rather than rote recall.
- Image Design: Comfortable identifying or creating visuals (e.g., graphs, diagrams) that are crucial to solving problems.
- AI Familiarity (Preferred): Experience working with LLMs like Gemini or ChatGPT is a plus.
Compensation
- Competitive rates based on the number and quality of questions submitted.
- Opportunity for ongoing collaboration as we expand the dataset and undertake future projects.
Why Join Us
- Help shape the next generation of AI by providing expert-level, original content.
- Contribute to a globally impactful project in AI research and STEM education.
- Collaborate remotely with a professional, motivated team.
- Enjoy flexibility and independence in your working hours.
Application Instructions
To apply via UpWork, please submit:
- Your resume or CV, emphasizing your academic background and subject expertise.
- A sample of prior academic work, such as original questions, assessments, or published content (if available).
- A brief note confirming your access to and ability to use Gemini 2.5 Pro or similar models.
Additional Notes
- This is a remote freelance role with flexible scheduling.
- All content will be reviewed and must pass originality and quality checks.
- Questions will undergo a review process to ensure they meet the dataset’s standards
Project Overview
Our goal is to build a challenging and novel dataset that pushes the boundaries of current AI capabilities in question answering. The questions must test deep conceptual understanding, multi-step reasoning, and problem-solving mathematics.
At least 50% of the questions must incorporate visuals (e.g., graphs, diagrams, models) that are essential to solving the problem. Each question will be tested against Gemini 2.5 Pro and must include a conversation log demonstrating the model’s performance with and without hints.
Challenges
- Questions must be challenging for current AI models, requiring expert-level knowledge and reasoning.
- Questions should be "Google-proof", meaning they cannot be easily answered through simple web searches.
- Answers must be unambiguous and verifiable to ensure consistency in AI evaluation.
Responsibilities
- Create high-difficulty questions (~graduate-level or upper undergraduate level) in your area of expertise.
- Write a detailed explanation and/or solution for each question.
- Ensure questions are original and not copied from textbooks or previous tests.
- Must be able to provide concise, informative hints designed to help advanced AI models reason through and answer the question correctly.
Qualifications
We are seeking individuals with the following profile:
- Education: A Master’s degree (or higher) in a STEM discipline is nice to have. Ph.D. holders are strongly preferred.
- Expertise: Deep academic or professional knowledge in mathematics, physics, chemistry, biology, or a related field.
- Writing Skills: Excellent written English; ability to write clear, rigorous questions and explanations.
- Reasoning Skills: Ability to craft problems requiring advanced, multi-step reasoning rather than rote recall.
- Image Design: Comfortable identifying or creating visuals (e.g., graphs, diagrams) that are crucial to solving problems.
- AI Familiarity (Preferred): Experience working with LLMs like Gemini or ChatGPT is a plus.
Compensation
- Competitive rates based on the number and quality of questions submitted.
- Opportunity for ongoing collaboration as we expand the dataset and undertake future projects.
Why Join Us
- Help shape the next generation of AI by providing expert-level, original content.
- Contribute to a globally impactful project in AI research and STEM education.
- Collaborate remotely with a professional, motivated team.
- Enjoy flexibility and independence in your working hours.
Application Instructions
To apply via UpWork, please submit:
- Your resume or CV, emphasizing your academic background and subject expertise.
- A sample of prior academic work, such as original questions, assessments, or published content (if available).
- A brief note confirming your access to and ability to use Gemini 2.5 Pro or similar models.
Additional Notes
- This is a remote freelance role with flexible scheduling.
- All content will be reviewed and must pass originality and quality checks.
- Questions will undergo a review process to ensure they meet the dataset’s standards