This project aims to address this gap by conducting a systematic, controlled study of human versus LLM-generated text detectability using paired question–answer datasets. Rather than proposing a novel detection architecture, the focus is on analyzing detection robustness, failure modes, and the impact of adversarial humanization strategies.

Experimental data and technical audit of LLM content detectors (GPT-5, Claude 4) vs. humanization algorithms.

Stock Bench is an LLM benchmarking system where LLMs compete in a prediction market, making bets on how well they’ll perform on tasks. Thus making it possible to measure each model's performance, as well as how accurate and self-aware each model is about their own performance.

LLM benchmark evaluation framework on Google Colab, this repository contains the full 18 week learning journey