Hello,
and welcome to my resume

Interface2 is a prototype agent-first desktop tuned for fully offline use on 4B-parameter models. You describe the task, see which helper is engaged, what it’s allowed to touch, and where the result will appear.

AI-DEOLOGY is my attempt at an impossible task. Though still in development, it already functions as a framework designed to analyze political bias in large language models from within the very systems it seeks to critique. The program enables users to select from both predefined and custom models, configure ideological test conditions, and evaluate responses across multiple controlled scenarios. The system centers on three main test types: evaluating raw responses to political questions, measuring influence after simulated biased conversations, and testing reactions to ideologically slanted prompts. A second model, acting as an impartial judge, analyzes the primary model’s outputs and rates their agreement with each question’s premise.

The framework supports both API-based and local models via a customizable configuration file. A streamlined command-line interface guides users through model selection, test execution, and result aggregation, offering an accessible yet rigorous environment for examining ideological patterns in language model behavior.

PropagandAI is an experimental platform and tool designed as an immersive environment for exploring the automation of news and ideological storytelling through artificial intelligence. Built around a few commercially available large language models, the system fetches, generates, and rewrites news content from across the ideological spectrum to reveal how bias, framing, and even entire narratives can be shaped at scale.

Through simulations, it exposes how AI can be used to fabricate consensus, polarize discourse, or mass-produce propaganda.