Job Title: AI Mathematics SME PhD / Adversarial Prompt Specialist

Location: Remote

Position Type: Freelance / Contract

About the Role

This role involves adversarial benchmarking and ground-truth dataset creation for advanced frontier AI models. Rather than standard grading or simple auditing, specialists design rigorous, original mathematical problems specifically intended to test, challenge, and stump state-of-the-art reasoning models.

What You Will Do (Exact Portal Workflow)

Adversarial Problem Creation: Draft complex, high-difficulty mathematical problems in the prompt interface and run live executions against frontier AI reasoning models to verify that the problem consistently stumps or breaks model logic.

Step-by-Step Solution Authoring: Write transparent, verified ground-truth solutions, breaking down the problem step-by-step into structured reasoning components using standard LaTeX syntax.

Solution Concept Tagging: Identify and summarize key mathematical principles or theorems applied in your solution into concise, single-line tags per strict platform guidelines.

Syntax & Formatting Compliance: Format final answers using strict platform constraints (e.g., matching mathematical expressions with Final Answer: boxed{}).

Automated Linter Debugging:



Utilize built-in evaluation tools to run real-time checks, address automated formatting error logs, and pass all system validation tests prior to task submission.

Candidate Requirements

Mandatory Qualifications

Degree (PhD) in Mathematics, Applied Mathematics, Physics, Statistics, or a related quantitative field.

Proven track record of academic contribution (at least one published paper, journal article, conference submission, or book chapter).

Proficiency with LaTeX for mathematical typesetting.

Strong written communication skills in English.

Preferred Experience

Experience with adversarial prompting, red-teaming, or AI model benchmarking.

Background in mathematical competition design (AMC/AIME/Olympiad level) or advanced academic content creation.

Familiarity with debugging string-matching validation rules and automated feedback tools.

Project Execution Details

Platform Environment: Interactive browser-based workflow platform with live AI model execution and automated linter integration.

Schedule: Fully flexible hours, 100% remote.

Tracks Available: Mathematical Logic (Non-Coding) and Computational Mathematics (Coding/Python).

📌 Mathematics SME | PhD (India)
🏢 PRISM Tech Systems
📍 India