I turn messy real-world problems into structured models, testable workflows, and working software.

My background is in automotive engineering and process/quality thinking. My current work explores how AI agents, local-first software, and evidence-based research can extend an engineer's ability to learn, decide, and build.

An evidence-guided Agent Skill for entering unfamiliar fields. It builds concept dependency maps, grades evidence, compares professional options, and runs mastery and transfer loops instead of producing generic reading lists.

Agent Skills · research systems · decision design · mastery learning

An Obsidian-based content workflow distributed as a working plugin release. It focuses on turning fragmented material into a repeatable local knowledge and production system.

JavaScript · Obsidian · local-first

• LumaShelf — a private macOS media-library and playback application, developed in Swift with a strong automated-test suite.
• Project Radar — a local TypeScript dashboard that discovers and organizes development projects on the machine.
• Engineering agent workflows — reusable systems for automotive process, quality analysis, research, reporting, and knowledge transfer.

map the system → inspect evidence → define criteria
→ build the smallest real artifact → test → revise → transfer

I care about:

• Systems thinking over isolated features
• Evidence and explicit assumptions over confident summaries
• Local-first ownership and recoverable workflows
• Tests, rubrics, and feedback loops over “looks finished”
• Tools that strengthen human judgment instead of hiding it

Swift · TypeScript · JavaScript · Python · Git · AI agents

Keep creating new stuff — but make each project prove something.