Projects

A methodology for structuring AI-assisted development. Explores how systems thinking guides human-AI collaboration in software engineering.

Investigates whether local LLMs can drive a complete content generation pipeline. RSS ingestion, RAG-based writing, and automated publishing on local hardware.

Studies the interaction between local inference and everyday user workflows. A concrete experiment in unobtrusive on-device AI utility.

Explores the practical limits of local LLMs for content analysis. Built during the Loco Local LocalLLaMa Hackathon.

Examines local text-to-speech pipelines as a component in multimedia content creation. What does fully offline content generation look like?

Investigates whether agents can maintain persistent, self-evolving memory. A seven-layer cognitive architecture for autonomous knowledge synthesis — no cloud required.

Explores mixture-of-experts routing through dynamically selected personas. Can locally-hosted, specialized models compose into synthetic intelligence?

Examines GraphRAG as a retrieval architecture for document analysis. Compares graph-based reasoning against flat vector search.