Founder, Chedai LLC. Comprehensive Designer: publishing systems, dataset construction, and human–AI workflow architecture as one integrated practice.
I build working systems with large language models and document them so the work holds up to scrutiny. Local inference, deterministic validation, measurement before claims.
Focus areas
Local-inference publishing and dataset pipelines Deterministic validation of model output Human–AI workflows that can be understood and audited Methods that prioritize clarity, traceability, and durability
- LLM Export Analytics: tooling for analyzing ChatGPT and Claude export data, built with privacy, measurement, and reproducibility in mind. Includes a published methodology.
- VCAT: a structured-data toolkit for the Voynich Manuscript, with automated test coverage. Published on GitHub and Hugging Face.
- The Muser: an open-source natural-language music composition system in Python that orchestrates multiple AI models through an agent loop. MIT licensed.
- Clarity over cleverness
- Measurement before claims
- Documentation as a first-class artifact
- Build for durability, not visibility