LLM Reasoning Benchmark Questions + Python Evaluation Script

I need a freelancer to prepare benchmark questions and answers for testing a custom LLM’s reasoning ability.

Scope:
Question Set:
Collect 500–600 LLM benchmark questions with correct answers.
Focus areas: logical, mathematical, commonsense, analytical, and multi-step reasoning.
Deliver as JSON or CSV.
Python Script:
Load questions and send them to an LLM (I'll handle API integration).
Compare model answers to correct ones.
Output a simple accuracy report.
Requirements:
Knowledge of LLMs, reasoning datasets, or NLP is preferred.
Clean, documented code.
Use only open or original questions.