CTO & Principal AI Architect. I build agentic AI, RAG and voice AI systems that run in production, and lead the teams that keep them running.
Most of it runs where mistakes are expensive, in healthcare, legal and finance, with the checks that let a business trust it in production.
- 30+ AI/ML systems in production, and 50+ built, in the past three years
- 2,600+ pull requests and 14,000+ commits across 28 repositories since October 2024
- CTO & Partner at Pendoah, and fractional CTO to startups that have each passed $1M ARR
- Former CIO of Perry Homes while revenue grew from $800M to $1.71B
- 23+ years in technology leadership · Houston, TX
| Role | What it involves |
|---|---|
| CTO & Partner, Pendoah (Oct 2023 to present) | Oversight across a 110-person firm. I directly lead the 20+ people in its software and AI group. |
| Fractional CTO | CTO to startups, each past $1M ARR. At one $1M+ ARR AI SaaS startup I am fractional CTO and Head of Product Engineering and Forward Deployed Engineering: a five-person team that shipped 500+ pull requests in about eight weeks. |
| CIO, Perry Homes (Jun 2017 to Feb 2022) | Led a 45-person technology department plus a 30+ extended team for a homebuilder of about 1,300 employees and 2,500 contractors and suppliers. Retired about 70% of legacy applications and cut IT opex about 20% while revenue grew from $800M to $1.71B. |
| Earlier | Principal at Credera · Director of Enterprise Architecture at PwC · Practice Director at Hitachi Consulting · Director of Technology Strategy at Accenture |
| Product | What it does |
|---|---|
| ABACare | Helps ABA therapy clinics get paid for every session they deliver, from authorization to claim. Preparing design-partner pilots. |
| Tellgence (built at Pendoah) | Voice AI that answers a small business's phone, books the appointment and hands off to a person. |
| Project | Status | What it does |
|---|---|---|
| AgentHub | Active | Keeps every AI coding agent on your machine (Claude Code, Codex, Cursor, Gemini CLI, Copilot CLI and more) running the same skills, MCP servers and policy, and tells you when one drifts. Apache-2.0. |
| fakeplane | On npm | Test AI agents against realistic fake support desks, Slack and Linear, with real-world failures, before they touch production. Apache-2.0. |
| replyonce | On npm and PyPI | Stop AI agents from sending a customer the same reply twice. For Node.js and Python. Apache-2.0. |
| subtext-scan | On npm | Catch hidden instructions in PDFs, Word files and web pages before an AI agent reads them. Apache-2.0. |
Multi-agent handoffs · agentic RAG · text-to-SQL agents · realtime voice agents · deep research · human-in-the-loop approval · MCP servers · deterministic guardrails · evals that gate every release