I build single-file HTML research platforms with local-first AI integration — applications that run entirely from one file, store data in IndexedDB, and put the user fully in control of their data and their AI provider choice.
| Project | What it does | Status |
|---|---|---|
| Rosetta HTM™ | Universal decipherment platform for undeciphered ancient scripts (Indus, Linear A, Proto-Elamite, +5 more). 39-paper Published Hypotheses Library. Three Egyptian "honesty tests" that detect AI cheating. | Public |
Several other research-grade single-file platforms exist as personal tools for retirement planning, estate organization, and family health records. Those are not currently published.
- Single file, no build step, no dependencies. Open the HTML, it works.
- Local-first. IndexedDB / localStorage; no servers; no telemetry.
- AI provider neutrality. Bring your own key — Anthropic, Groq, OpenAI, local Ollama. Switch at runtime.
- Honest limits. Where AI is involved, build mechanisms to detect AI hallucination (see the Egyptian Sabotage Test in Rosetta HTM™).
- Documentation as a first-class artifact. Every platform ships with a comprehensive PDF reference guide alongside the app.
- Attribution as a non-negotiable. When research depends on others' work, that work is acknowledged prominently and permanently.
Recently exited a federal contracting business after two decades. Now in the post-sale transition, building research-grade tools I'd want as a working scholar or practitioner.
- Cross-script decipherment platform research (Indus + Linear A + Proto-Elamite cross-validation)
- Writing The Calculus of Nations and developing its companion website
- AI integration patterns where local LLMs handle privacy-sensitive work and cloud APIs handle scale
- Self-hosted infrastructure for personal data sovereignty
For Rosetta HTM™ specifically — open an issue on the repo. For other inquiries, contact channels are listed in each project's README.