I am a wildly curious engineer that has always been drawn to the potential of technology.

What pulls me in is how every layer of machinery beneath a surface shapes what someone feels on top of it. I try to understand the real goal beneath whatever problem gets surfaced, and to keep hold of the humanity inside the technology.

• Knowledge Recall & Agent-Grounding System — a personal system that captures, synthesizes, and recalls knowledge to assemble grounding for an LLM. The recall layer is the part most "second brain" tools skip, a set of hot/warm/cold salience tiers with cued push and facet-based pull, exposed to agents through a set of MCP servers, with an eval harness that runs a headless agent through real tasks and checks it stayed on the intended path. It mixes local models with frontier ones to keep context lean.
• Seekbox — a save-and-recall product for articles, video, and podcasts. Browser extension, Next.js web app, and a GraphQL API, with an OpenGraph enrichment pipeline and weighted full-text search over Postgres.

Both are solo projects and private for now. I think out loud through the writing below.

• Core — TypeScript (front and back)
• Frontend — React/Solidjs/Vue, Next.js, Zustand, Astro, HTML, CSS
• Backend and data — Node.js, GraphQL, REST, Postgres, Drizzle, Supabase
• AI and infra — MCP servers, LLM eval harnesses, CI/CD
• Learning — Python, Bun, Deno

• Please be specific — the promise of AI is turning ambiguity into precision, and how you forfeit it by answering with more context instead of more structure
• Survival of Wonder — keeping AI an honest sounding board when it's built to earn your approval
• What makes us human — technology earning its place by clearing what's in the way

LinkedIn