I build with AI, then open-source what works. Mass-producing prototypes, one rabbit hole at a time.
I think the best way to learn AI is to ship with it. So that's what I do.
Right now I'm deep in the rabbit hole of spatial intelligence, world models, and embodied AI. If you're thinking about how machines learn to see, move, and reason about the physical world, we should talk.
ai-learning-engine — 32-feature adaptive learning engine with SM-2 spaced repetition, flex sessions, teach-back, and concept linking. Hub-and-spoke architecture. Built on 40+ academic sources in cognitive science. You bring a curriculum on any topic — the engine handles the pedagogy. Currently using it to study spatial AI and robotics. Works with Claude Code, OpenAI Codex, and Cursor AI.
context-cli — Audit any URL for AI crawler readiness. Scores content density, robots.txt, schema.org, and llms.txt on a 0-100 scale. Published on PyPI, includes MCP server.
| Project | What It Does |
|---|---|
| ai-learning-engine | Evidence-based adaptive learning engine — SM-2 spaced repetition, Socratic method, pre-testing, mastery gates. Any topic, any AI assistant. |
| context-cli | LLM readiness linter for websites (PyPI package) |
| investment-research-sdk | 79-agent multi-agent investment research system |
| DadCoach | AI parenting companion for new fathers (Android, Kotlin, Gemini) |
Currently using ai-learning-engine to learn Spatial AI from scratch. Day 4/50 | 38 terms tracked | 3 sessions complete
Building in public, shipping in private, open-sourcing when it works. Say hi on X/Twitter.