AI Agent Developer — building voice agents, MCP servers, and automation pipelines that actually work in production, not just in demos.
I design and deploy AI agent systems for real business use cases — from Model Context Protocol (MCP) servers that connect AI to real tools and data, to voice assistants that handle live phone calls, to automated workflows tying it all together. My focus is production-readiness: proper authentication, logging, monitoring, and security — not demos that only work once. I bring real domain background from the oil & shipping industry alongside hands-on AI agent development, through BloomtechAI Solutions.
A Model Context Protocol server exposing live tools to AI agents, migrated from local-only stdio transport to a cloud-deployed HTTP service on Render — with authenticated endpoints, request logging, uptime monitoring, and a clean dependency-security audit. Officially listed on the Model Context Protocol Registry, and approved and published on Glama (v1.0.0) — one of the largest MCP server directories, covering 80,000+ servers.
A voice assistant that answers real phone calls for a restaurant and books food orders — reliable message capture, tuned interruption/silence handling, automated call analytics, and an authenticated n8n data pipeline. Two real production bugs were found and fixed during development rather than papered over.
A voice AI agent for a real estate agency that has a natural conversation with callers, then determines whether they're a qualified prospect (real budget, timeline under 3 months, decision-maker) before flagging them for human follow-up. Verified correct in both directions via real test calls, including a real currency-assumption bug found and fixed.