Production LLM Systems · Enterprise Adoption · Agents, MCP & Developer Workflows
I build AI systems people can adopt, operate, and trust. Work spans AI enablement, enterprise deployment, agents, RAG evaluation, MCP security, and human-in-the-loop operations. Approach to safety is practical: test real failure modes, protect connected tools and data, preserve human judgment, own what happens after launch.
New York City · U.S. permanent work authorization · No sponsorship required · Open to NYC and U.S. remote leadership roles
mcp-scan: open-source TypeScript security scanner for Model Context Protocol configs. Catches tool poisoning, prompt injection, credential leakage, data exfiltration, and supply-chain risk. On npm, SARIF 2.1.0 output, Homebrew tap. npx mcp-scan@latest
ProTeach: homeschool platform where certified teachers craft weekly lessons per child, with 17 learning games, realtime chat, and portals for parents, kids, and teachers.
rag-eval-toolkit: Python toolkit for evaluating RAG pipelines with LLM-as-judge scoring, retrieval metrics, and hallucination checks. On PyPI, MIT. pip install rag-eval-toolkit
ThynkQ: AI engineering studio for MVP development, AI integration, and fractional CTO work.
| Company | Role | Impact |
|---|---|---|
| PowerLiens | Head of AI & Engineering | Own product architecture and production systems for a legal-operations platform: CRM, intake, provider search, case tracking; building AI-assisted intake automation. |
| Meta (contract) | AI Trust & Safety Analyst | Red-teamed LLaMA models through systematic adversarial testing. |
TypeScript · Next.js · React · Vercel · Firebase · Stripe · Python · Docker · CI/CD
- Agency: thynkq.com
- Email: abanoub.rodolf@gmail.com

