I build AI integrations for companies that need more than a proof of concept. Chatbots that actually understand your domain. RAG pipelines that pull from your real documents. Autonomous agents that connect to your tools and data through MCP servers.
I've been shipping software for over 20 years. Started in Java, moved to Go and Python, spent the last several years working in cloud architecture across AWS, GCP, and Azure. The AI work is the latest thing I've added, and it sits on top of everything else I know. I bring an engineering mindset to projects that usually stall at the demo stage.
What I actually do day to day:
- Custom AI agents using Google ADK, LangChain, and MCP. Not wrappers around a single API call. Agents that reason about clinical workflows, check drug interactions, and escalate to humans when the stakes are high.
- RAG pipelines with pgvector and Pinecone. Document processing, chunking strategies, embedding optimization. The boring parts that make retrieval actually work.
- LLM integration with OpenAI, Claude, Gemini, and self-hosted models via vLLM. Including the parts nobody talks about: prompt engineering that holds up under real load, fallback chains, cost monitoring.
- Healthcare-specific work: FHIR R4 integration, HIPAA-compliant architecture, clinical decision support. Built for real medical SaaS platforms processing thousands of requests a day.
- Infrastructure: Terraform, Docker, CI/CD, serverless. The stuff that keeps your system running at 3am.
My stack: Python (FastAPI), Go (Gin/GORM), TypeScript/React, PostgreSQL, BigQuery.
Google Cloud certified (Professional Cloud Architect and Data Engineer). 15+ years freelance consulting, which means I've learned to communicate clearly and deliver, or I don't get hired again.
If you have an AI project that needs to work in production, not just in a demo, I'd love to hear about it.