RAG pipelines, agentic assistants, and Claude & OpenAI API integrations — built on 25 years of production enterprise architecture and a physics brain.
I'm a physics-educated software engineer who builds LLM-integrated systems — retrieval-augmented pipelines, multi-agent assistants, and Claude/OpenAI API integrations — on top of two and a half decades architecting enterprise-scale software for Fortune 500 insurers and defense programs.
Most "AI integration" is a ChatGPT call bolted onto a form. I build the real thing: stateful, multi-agent systems with vector retrieval, tool/function calling, streaming, and error handling that holds up in production. The engineering judgment comes from 25 years of building systems that can't fall down — the AI is just the newest layer on a very deep stack.
I'll tell you what's possible and what isn't, up front. No sales pitch, no hand-waving — straight answers and code you can actually maintain.
Claude or OpenAI wired into your app or workflow — streaming, tool calling, retries, secure key handling. Done right, not fragile.
Assistants that answer from your data — vector retrieval over your docs, grounded and accurate, not hallucinating.
Multi-step agentic workflows with tool/MCP integration — systems that do work, not just chat.
The plumbing AI runs on — Java, Python, Node, SQL/NoSQL, Docker, CI/CD, DevSecOps. 25 years of it.
Aging enterprise systems dragged into the present — containerization, cloud migration, CI/CD, security remediation. I've done exactly this at scale.
Not sure how to add AI to your product — or whether what you built is sound? I'll architect the integration or audit what you've got. Senior judgment, no guesswork.
A multi-agent assistant built from scratch with a temporal knowledge graph (thousands of nodes), retrieval-augmented context injection, Claude API orchestration, and MCP tool-servers running as always-on background services. Stateful memory that persists and retrieves across sessions.
A live production bot with Claude API integration handling real-time community interaction — conversational AI wired into a Node.js service.
Lead Architect over ~30 Group Life applications at a Fortune 500 insurer: Azure DevOps CI/CD pipelines, containerized delivery to OpenShift, a .NET Core rules engine replacing 1,000+ manual rules, and org-wide security remediation (Log4Shell, WebSphere migrations).
Architecture and hands-on development across an enterprise Group Life portfolio — CI/CD, containerization, rules engines, high-throughput call-center systems processing thousands of transactions per minute.
High-reliability defense software: Kalman-filter tracking algorithms for the Ballistic Missile Defense System, multi-threaded C++ on Linux clusters, PKI security modules, Windows-to-Linux ports.
Graduate-level math and physics — the foundation under the embeddings, retrieval, and model behavior. B.S. Mathematics, Stockton University.
Building something with AI and want it done right? Let's talk about it.