AI Systems Architect
Strategist. Builder. Visionary.
I design and ship production AI systems. Not proofs of concept — deployed, autonomous pipelines that handle real workloads without intervention. From compiler design to payment automation, every system I build is constraint-first: working within platform limits, API quotas, and hardware restrictions rather than fighting them.
12+ years of shipping production systems. 10 deployed and running. 6 Claude API integrations in production. I use Claude Code as my primary development environment daily — custom skills, MCP server integrations, and agent-driven workflows. This isn't a technology I'm experimenting with. It's how I build.
My work spans the full depth of software engineering: language implementation (compiler from scratch in TypeScript), desktop applications (Tauri + Rust), full-stack platforms (Express.js + Supabase), autonomous content pipelines (n8n + AI), and payment/subscription systems (PayPal + Discord). The common thread is systems thinking — every project is designed as an integrated system, not isolated features.
AI & Claude Integration
AI & Automation
Systems & Languages
Platform & Infrastructure
How I Work With AI
Claude as Primary Dev Tool
Every project in this portfolio was built with Claude Code as the primary development environment. Not as an assistant — as the tool I think through. Custom skills encode my engineering methodology into repeatable processes.
MCP & Tool Integration
Production MCP server integrations: Desktop Commander for system ops, n8n for workflow automation, custom servers for project-specific tooling. Claude interacts with external systems as a first-class capability.
Agent Architecture
Implemented all 7 patterns from Anthropic's agent guide. Built FORGE (pre-authorization policy engine for autonomous AI agents) and Hey Jimmy (multi-channel AI command routing). This is where AI engineering is heading.
The Journey
Started building hardware scripting systems for specialized controllers. Constraint-driven development from day one — 2KB memory, 50ms execution frames.
Expanded into AI integration. Built the first Claude API integration into a Tauri desktop app with knowledge base injection and dynamic prompt assembly.
Shipped autonomous content pipelines (180+ videos/month), full-stack platforms with Claude API, and payment automation systems. Adopted Claude Code as primary dev tool.
Built a compiler from scratch in TypeScript. Shipped .files (Tauri v2 desktop app), a CRM system, FORGE (AI task runner with policy engine), and deep MCP/agent pattern work. 10 production systems, 6 Claude integrations.
Philosophy
1. Correctness before impressiveness
Software must behave correctly before it looks impressive.
2. Constraint-first engineering
Platform policies, API limits, and technical constraints are engineering inputs, not obstacles. The best solutions work within constraints rather than fighting them.
3. Production-first mindset
All systems built with production requirements from day one: logging, error monitoring, automatic recovery, graceful degradation.
4. AI as a system component
Not a novelty — a tool with defined inputs, outputs, and failure modes like any other system component. Integrated into architecture, not appended to it.
5. Ship, then iterate
A deployed system teaches more in a week than a planned system teaches in a month. Get it into production, then refine based on real behavior.