Angel AI is the project I am building to bring my engineering skills together into one substantial application.
It is designed around:
- 🧠 Local AI — local Ollama inference and replaceable local models
- 💾 Persistence — durable local state and SQLite-backed data
- 🛡️ Safe automation — controlled tools, validation, permissions, and bounded execution
- 🔄 Recovery — portable backups, verification, and recovery workflows
- 🧪 Testing — automated regression and cross-machine validation
- 🪟 Windows engineering — Windows 10/11 application and administration workflows
- 📚 Continuity — architecture, decisions, handoffs, contributor assignments, and testing guidance in
rusty/
Genesis 1.0 — v1.0.0
Next work in progress: improving the Current Information / WeatherBrain pipeline so Angel reliably routes date/time and weather requests through the intended capabilities instead of unnecessarily falling back to general web search.
| Engineering Area | What I build |
|---|---|
| AI Applications | Local AI assistants, capability routing, context and memory |
| Windows | Desktop software, PowerShell administration, automation |
| Systems Integration | APIs, browsers, local services, external tools |
| Reliability | Backups, recovery, validation, persistence |
| Automation | Repeatable build, test, administration, and recovery workflows |
| Testing | Regression testing, environment testing, evidence-based validation |
| Documentation | Architecture, decisions, handoffs, contributor continuity |
| Git / GitHub | Checkpoints, releases, protected history, collaborative workflows |
Protect what works. Inspect before changing. Build in modules. Test what matters. Document what was learned. Make recovery possible.
I don't want software that only works on the developer's machine.
I want software that can be understood, tested, recovered, continued, and improved by the next person.
Angel AI is being developed as a real engineering project, not just a demonstration.
If you want to contribute, test Angel on different hardware or Windows environments, investigate an issue, improve documentation, or have your own AI assistant help with the project, start with the rusty/ continuity documentation in the repository.
The goal is simple: don't duplicate work — leave better evidence for the next person.
