Independent builder shipping AI-native tools for developer workflows, sports intelligence, and creative production. I mostly work in TypeScript, Python, and Next.js.
My current attention is on harnessing agents and AI models: making them observable, composable, and useful inside real building workflows.
Current focus: agent harnesses, model workflows, Sports AI surfaces, and creative production systems.
| Project | What it does | Stack |
|---|---|---|
| awesome-sports-ai | Curated sports-AI tools, models, data APIs, and analytics resources. | JavaScript |
| QuickFork | Turns GitHub repositories into fast overviews, launch copy, and share-ready infographic assets. | TypeScript |
| DevLog | Searchable work journal and dashboard for coding-agent sessions. | TypeScript |
| Project | What it explores |
|---|---|
| videoclaw | Agent OS for AI video generation, from planning and model selection to audit-ready outputs. |
| DevLog | Searchable work journal and dashboard for coding-agent sessions. |
| vibe-producing | Browser music creation from hummed melodies into loops and lyrics. |
- I care about agent-native software that produces inspectable artifacts, not black-box demos.
- I like model workflows that can be prompted, audited, retried, and reused.
- I like tools that turn messy creative, sports, or coding work into repeatable, reviewable systems.
- Most experiments start small, ship quickly, and get tightened through real usage.