I build AI systems that turnrepetitive work intoreliable automation.

Designed and built a single lead pipeline for a US-based home-services company, replacing scattered tools with five coordinated n8n workflows. Leads from three WordPress form paths are normalized, duplicate-checked, and synced to both GorillaDesk and a Google Sheets CRM with a concise summary of what the customer needs. An estimating layer applies configurable service-area and pricing rules, routing anything uncertain to manual review instead of guessing. Staff notifications run over Gmail and Twilio SMS with an acknowledgement reminder loop, and every estimate override is captured in an append-only audit trail that syncs idempotently back to the CRM.

Built a web interface where users input location and business type, then the system automatically searches the internet for matching businesses, collects data, and saves structured results to Google Sheets. An n8n workflow processes the data for enrichment and follow-up. Fully automated from user input to organized lead database.

Intelligent email classification and routing system using Claude API with tool use, SSE streaming, and n8n integration for automated workflow execution. Processes incoming emails, classifies intent, and routes to appropriate handlers.

End-to-end multi-agent pipeline: email intake → Claude orchestration → sub-agents → tool use → structured output to Sheets/DB. Features SSE streaming and FastMCP for standardized tool access.

Built a Model Context Protocol server for standardized AI tool access with SQLite backend, Flask REST API, and CSV data import. Enables any MCP-compatible AI client to query and manage structured data.

Built an end-to-end lead management workflow using n8n, Groq AI, Google Sheets, Gmail, and Telegram. The system automatically qualifies incoming leads, scores buying intent, recommends services, sends personalized follow-ups, and notifies the business owner in real time.