Senior Project Manager turned AI Operations engineer. I build production agentic systems that solve real business problems.
Most of my work lives in private repositories (company IP), but here's the shape of it:
Production Agentic Automation
- Designed an autonomous AI agent that processes enterprise support tickets end-to-end: classifying, analyzing, deciding, and executing actions through third-party APIs
- Built the webhook orchestration layer, confidence gating, and approval workflows that make autonomous operation safe for production
Multi-Service Platform Engineering
- Architecting a full-stack operations platform: FastAPI backend, React frontend, PostgreSQL, Redis, OAuth/OIDC auth, WebSocket real-time updates
- 7-service Docker Compose environment with department-level dashboards, agent monitoring, audit trails, and role-based access control
- Designed for production deployment on AWS (EC2, ALB, Secrets Manager)
Local AI Research Lab
- Dedicated hardware for fine-tuning and local inference experiments
- Raspberry Pi cluster for hosting open-source tools, vector databases, and test environments
- Exploring fine-tuning economics, RAG architectures, and edge AI deployment
I use AI agents as implementation partners, and I'm direct about that. Here's what that actually looks like:
- I architect systems, write technical specifications, and design the overall structure
- I use spec-driven development (SDD) and test-driven development (TDD) to define what "done" means before any code gets written
- AI coding agents (Claude Code, Kiro) implement against those specs while I review, course-correct, and handle the integration work
- I maintain the feedback loops, steering documents, and quality gates that keep AI-generated code production-grade
I understand the code, the architecture, and the tradeoffs while AI handles velocity. I handle direction, quality, and accountability.
Languages: Python (primary), JavaScript/TypeScript (frontend) Backend: FastAPI, PostgreSQL, Redis, Alembic, WebSockets Frontend: React, TailwindCSS, Vite Infrastructure: Docker Compose, AWS (EC2, ALB, Secrets Manager), Authentik (OAuth/OIDC) AI/ML: Anthropic Claude, Google Gemini, structured output, model cascading, confidence scoring Dev Tools: Claude Code (Opus), Kiro IDE, uv, Git worktrees, Linear Hardware: Apple Silicon (128GB unified), Raspberry Pi cluster, 3D printer (Bambu Labs X1C)