I build practical tools for local AI, creative workflows, and everyday automation.
My focus is making workflows usable and recoverable: repeatable setup, clear diagnostics, persistent background work, and checks that exercise failure cases. I’m interested in AI implementation, AI operations, evaluation, and business automation roles.
| Project | Problem and implementation | Evidence |
|---|---|---|
| llama-wrap | Run and troubleshoot local GGUF model servers through a shared GUI/CLI core: saved configurations, endpoint probes, tool-call checks, benchmarks, and context stress tests. | Implementation case study · Tests · Releases |
| Papan + ComfyUI | Collect references locally, extract DWPose control videos, and connect image/video references to ComfyUI through a companion node. | Implementation case study · ComfyUI extension · Verification |
| Papan for Android | Collect references on a phone and deliver links to a paired desktop, with destination selection and queued retries. | Source and verification · Releases |
Desegnos — An AI-assisted visual design desktop app with an infinite canvas, scoped AI editing, and local project history. Export paths support websites, slides, and images.
Built with React, TypeScript, Rust, and Tauri. In active development, with end-to-end editing and export validation underway. The repository is private while development continues.
I use AI-assisted development and document the implementation through source code, tests, release records, and clear limitations.
