Peace M.

I build AI agents, LLM apps, and automations, and I secure them so you can ship AI you actually trust. I find what breaks them, prompt injection, data leakage, over-permissioned agents, and I fix it before an attacker finds it first, AI Red teaming, threat modelling. Most teams ship an AI agent that works great in the demo, then realize it can read every customer record and act on a stranger's hidden instruction buried in a document. That is the gap I close. WHAT I BUILD - AI voice and chat agents that answer, qualify leads, book calls, and hand off to a human when it matters - LLM and RAG apps that answer from your own data, with sources, so they do not make things up - Invoice and document intelligence that catches errors, duplicates, and fraud - Automations across Make, n8n, Zapier, and GoHighLevel that remove the repetitive work HOW I SECURE IT - Red teaming AI systems against the OWASP LLM Top 10 (prompt injection, data leakage, insecure output, excessive agency) - Threat modeling with STRIDE and data-flow diagrams, mapped to MITRE ATLAS - Guardrails, least-privilege agents, input validation, and human-in-the-loop controls - AI governance and responsible-use policies aligned to NIST AI RMF and ISO 42001 I am CompTIA SecAI+ certified, so the security side is grounded in current standards.. TOOLS: Python, FastAPI, React, LangChain, PydanticAI, Google Cloud (Cloud Run, Document AI), Pipecat, Twilio, Deepgram, ElevenLabs, PostgreSQL, Docker, GoHighLevel, Make, n8n, garak. promptfoo If you are putting AI in front of real customers and real money, let us make sure it is safe to ship. Send me a message and tell me what you are building.