On sabbatical after 10.5 years at Amazon. Exploring epistemology — what humans should know vs. what we delegate to AI.
Most AI tooling asks: "How do we make AI do more?"
I'm more interested in: "What should humans still do — and why?"
This is applied epistemology. Not philosophy for its own sake, but building tools that make the human-AI boundary explicit and intentional.
Autonomous spec-driven development orchestrator for Claude Code. One word (SDLC) triggers the full lifecycle — requirements, design, tasks, implementation — with dual-critic validation (Advocate/Skeptic) and intelligent task batching. Built on a neuroscience insight: the agent that writes code should never review it.
Config-driven hook framework for Claude Code. Guard rails, coaching, analytics, and an interactive TUI — all from one YAML file. 1,300+ tests, 12 built-in recipes, daemon feed platform with 5 producers and 19 status line segments.
|
MCP server for spaced repetition practice tracking. 13 tools for mastery evidence, gotcha patterns, study plans, and rep logging. Built on proven interval scheduling with intelligent review surfacing.
|
Multi-agent mock interview transcript analyzer with anti-hallucination protocol. 4 parallel agents (Strengths, Mistakes, Behavioral, Factual) produce confidence-scored insights across 10 categories.
|
|
Observability dashboard for Claude Code. Track MCP server utilization, health, and errors. Zero-daemon architecture with SQLite WAL. |
MCP server for Splitwise. Automates expense splitting with duplicate prevention and deterministic math verification. |
|
Claude Code skill that transforms tutorials into interview-ready flashcards. Quizlet-compatible output with reasoning-focused Q&A.
|
OpenAPI 3.1 specification for "Designing Your Life" methodology. AI-first life coaching framework with 13 entities, 6 rule sets, and prompt-aware extensions.
|
- google-calendar-mcp#157 — Day-of-week fields to fix LLM date hallucinations
- lazy-mcp — Lazy-loading for reduced context overhead
- mcp-server-asana — Custom tool discovery patterns
10.5 years at Amazon — Tech Lead building voice interfaces, communication APIs, and enterprise integrations at scale. Learned how to ship reliable systems to millions of users.
Now: Stepping back to think about where this is all going. Reading epistemology. Building tools. Figuring out what kind of work I want to do next.
Areas of focus
| Domain | Experience |
|---|---|
| AI Systems | LLM orchestration, agentic frameworks, RAG pipelines, context engineering |
| Backend | Go, Python, TypeScript, Java |
| Infrastructure | AWS, distributed systems, event-driven architecture |
- Building pdlc-autopilot — autonomous multi-agent development orchestration
- Building hookwise — config-driven guardrails and coaching for agentic coding
- Building tools for agentic coding workflows (MCP servers, eval frameworks, spaced repetition)
- Interviewing for Staff+ roles in AI infrastructure and developer experience
