Steven J. Hubbard · Solutions Architect · remote nationwide
About 18 months putting LLM tooling into production QA, code review and documentation, not demos. Claude Code and ChatGPT are daily engineering methods, backed by custom agent skills, purpose-built subagents and MCP integrations, written by someone who also ships the application around them.
LLM integrationTool and function calling, retrieval-augmented generation (RAG) over client documentation, and structured prompt evaluation.
AI-assisted deliveryAI-assisted QA, code review and documentation embedded in real sprint cycles, with measurable results rather than demos.
Agent systemsClaude Code and ChatGPT as daily engineering methods: custom agent skills, 17 purpose-built subagents, MCP integrations with Gmail, Google Drive, Google Calendar and a Markdown knowledge base, and scheduled agent pipelines with automated verification gates.
Internal toolingInternal tools built with Retool and Python, and the automation around them.
Integrated production LLM tooling into client delivery workflows: tool and function calling, retrieval-augmented generation over client documentation, and structured prompt evaluation driving AI-assisted QA and code review.
The platform behind an AI-powered Rate-to-Reveal game and the proprietary EyeQ+ scoring metric, with automated QA and data validation in the delivery pipeline that cut manual development effort 30%.
Integrated AI-driven analytics that boosted campaign precision 15%.
Available now for contract, contract-to-hire, fractional and full-time work. I usually reply within a day.