XAI 2.0

Explainable Autonomous Systems

Graduate research exploring how Explainable AI and symbolic reasoning can make autonomous AI systems more transparent, trustworthy, and easier to understand.

Symbolic Storytelling

Symbolic Approach to Fine Tuning LLM

Examines the integration of symbolic AI and large language models for interactive storytelling, demonstrating how structured metadata, retrieval-augmented generation, and rule-based reasoning improve narrative consistency in AI-driven game systems.

FLUX Isn’t Magic

A guide to the quirks of Flux

A practical look at building reliable AI image-generation workflows, sharing lessons learned from more than 500 FLUX-generated portraits, including prompt engineering, reproducibility, performance tuning, and creative best practices.

Forging the Future

The story behind the architecture that transformed Forge42Engine from a game project into a hybrid AI research platform

This article explores the design philosophy behind Forge42Engine, explaining how symbolic reasoning, deterministic rules, semantic memory, and large language models work together to create believable AI-driven worlds. It also traces the project’s evolution from an academic experiment into the research platform that now powers DM42, Tharendell, and my graduate work in hybrid AI and explainable systems.

Part 1: How RAG Turns NPCs Into Agents that Remember

Why NPCs forget everything, and how Retrieval-Augmented Generation gives them a past, a history, and a place in your world.

Part 2: Giving NPCs World Knowledge Through RAG

How I rebuilt canonical world knowledge as a searchable semantic layer, and gave NPCs the power to speak truthfully about a world they

Part 3: How RAG Enables Multi-Campaign Storytelling

How we turned retrieval into a dynamic narrative engine, and taught NPCs to guide players through branching worlds and evolving quests.

Advanced RAG Theory

Embedding Dynamics, Retrieval Policy, and the Emergence of Multi-Agent Cognition

Part 1: Why I Give Every NPC their own AI Model

A practical guide to identity-stable AI characters, built for worldbuilders, game developers, and fellow mad scientists.

Part 2: Seed Generation – Building Situational Intelligence

How structured synthetic seeds become the bedrock of identity-stable AI characters.

Part 3: Response Generation – Teaching the NPC

How to turn structured seeds into rich, persona-consistent long-form data.

Part 4: Fine-Tuning & Deployment – Forging the Brain

How 750 synthetic persona samples become a fully fine-tuned, identity-locked 3B LLM served from Python on RunPod.